data_preprocessing.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Control Parameters "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The lab_black extension is already loaded. To reload it, use:\n",
      "  %reload_ext lab_black\n",
      "The pycodestyle_magic extension is already loaded. To reload it, use:\n",
      "  %reload_ext pycodestyle_magic\n"
     ]
    }
   ],
   "source": [
    "%load_ext lab_black\n",
    "%load_ext pycodestyle_magic\n",
    "%flake8_on -i E203,E501 -m 119"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import datetime\n",
    "import os\n",
    "import sys\n",
    "import time\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import copy\n",
    "from joblib import dump, load\n",
    "import unittest\n",
    "from pandas.util.testing import assert_frame_equal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Msg_structure:\n",
    "    def __init__(self, name=None, feature_list=None, feature_rename_list=None):\n",
    "        self.name = name\n",
    "        self.feature_list = feature_list\n",
    "        self.feature_rename_list = feature_rename_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Modify This"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_sim_not_real = True\n",
    "policy_xy_not_z = True\n",
    "mix_sim_real_data = False\n",
    "\n",
    "if policy_xy_not_z:\n",
    "    SimResults_path = \"./blimpSim/results_XY\"\n",
    "    RealResults_path = \"./blimpReal/results_XY\"\n",
    "    MixResults_path = \"./blimpSim/mix_results_XY\"\n",
    "else:\n",
    "    SimResults_path = \"./blimpSim/results_Z\"\n",
    "    RealResults_path = \"./blimpReal/results_Z\"\n",
    "    MixResults_path = \"./blimpSim/mix_results_Z\"\n",
    "\n",
    "if data_sim_not_real:\n",
    "    data_folder_path = \"./blimpSim/data/behaviour_policy\"\n",
    "    exp_index = [\n",
    "        \"coil\",\n",
    "        \"double_coil\",\n",
    "        #         \"reverse_coil\",\n",
    "        #         \"reverse_double_coil\",\n",
    "        #         \"down_coil\",\n",
    "        #         \"reverse_down_coil\",\n",
    "        #         \"square\",\n",
    "        #         \"double_square\",\n",
    "        #         \"lines\",\n",
    "        #         \"lines2\",\n",
    "        #         \"reverse_lines\",\n",
    "        #         \"reverse_lines2\",\n",
    "        #         \"manual\",\n",
    "    ]\n",
    "\n",
    "else:\n",
    "    data_folder_path = \"./blimpReal/data/behaviour_policy\"\n",
    "    exp_index = [\n",
    "        \"rb1_201208_113634_6ca016cb2a35d1893682c089b7e118cfe3376122\",\n",
    "        \"rb1_201214_123513_6ca016cb2a35d1893682c089b7e118cfe3376122\",\n",
    "        \"rb2_201208_120334_747cb38938c76c535739e377d7ddcf7941929cb6\",\n",
    "        \"rb2_201214_124645_1fb28b582a0c07b1e0c355c96faae3a309a6a028\",\n",
    "        \"rb3_201208_125205_480c461174624e388ace9ac04608f91aba1ad15c\",\n",
    "        \"rb3_201214_125422_23f66a6fd5be1805bf7cd07b0b1bf6655c6d6fe2\",\n",
    "        \"rb4_201208_130611_927a9fcc88f638d5794133d62908a81779df33dd\",\n",
    "        \"rb4_201214_134747_ba0a89de34290b12aeac7a1962bb37d0b15bb022\",\n",
    "        \"rb5_201208_132010_c1b0c71b7bf8e4fd28cd75e5b7e3015a2d19ed23\",\n",
    "        #                     'rb5_201214_160837_2d7233b5a264488155044c3ef7380fa44a54c7ef', # missing several msgs\n",
    "        #                     'rb6_201208_153458_50f011aee1008fd7935e8590229482234626fb28', # missing several msgs\n",
    "        \"rb8_201208_182852_e26f0b4801296bc4b9e1a98870333bf47e1925ef\",\n",
    "        \"rb9_201208_164241_4bc6e0115172ea94797ea0ce08fcc639b781de0a\",\n",
    "        \"rb10_201208_164242_f3252431ffc58f5fc6b41dcf531f159d9b5f6221\",\n",
    "        \"rb11_201208_164241_b03cfd1b1cf180055bc0bc2e3d0eee0e73435cbe\",\n",
    "    ]\n",
    "\n",
    "\n",
    "if data_sim_not_real:\n",
    "    name_space = \"blimp-\"\n",
    "    Imu_msg = Msg_structure(\n",
    "        name=name_space + \"tail-imu\",\n",
    "        feature_list=[\n",
    "            \"Time\",\n",
    "            \"linear_acceleration.x\",\n",
    "            \"linear_acceleration.y\",\n",
    "            \"linear_acceleration.z\",\n",
    "        ],\n",
    "        feature_rename_list=[\n",
    "            \"Time\",\n",
    "            \"linear_acceleration.x\",\n",
    "            \"linear_acceleration.y\",\n",
    "            \"linear_acceleration.z\",\n",
    "        ],\n",
    "    )\n",
    "\n",
    "else:\n",
    "    name_space = \"machine_-\"\n",
    "    Imu_msg = Msg_structure(\n",
    "        name=name_space + \"Imu\",\n",
    "        feature_list=[\n",
    "            \"Time\",\n",
    "            \"linear_acceleration.x\",\n",
    "            \"linear_acceleration.y\",\n",
    "            \"linear_acceleration.z\",\n",
    "        ],\n",
    "        feature_rename_list=[\n",
    "            \"Time\",\n",
    "            \"linear_acceleration.x\",\n",
    "            \"linear_acceleration.y\",\n",
    "            \"linear_acceleration.z\",\n",
    "        ],\n",
    "    )\n",
    "    transmitter_msg = Msg_structure(\n",
    "        name=name_space + \"Transmitter\",\n",
    "        feature_list=[\"Time\", \"ROSControlled\", \"Armed\", \"Thrust\"],\n",
    "        feature_rename_list=[\"Time\", \"ROSControlled\", \"Armed\", \"Thrust\"],\n",
    "    )\n",
    "\n",
    "orientation_msg = Msg_structure(\n",
    "    name=name_space + \"pose\",\n",
    "    feature_list=[\n",
    "        \"Time\",\n",
    "        \"orientation.x\",\n",
    "        \"orientation.y\",\n",
    "        \"orientation.z\",\n",
    "        \"orientation.w\",\n",
    "        \"angVelocity.x\",\n",
    "        \"angVelocity.y\",\n",
    "        \"angVelocity.z\",\n",
    "        \"velocity.x\",\n",
    "        \"velocity.y\",\n",
    "        \"velocity.z\",\n",
    "    ],\n",
    "    feature_rename_list=[\n",
    "        \"Time\",\n",
    "        \"orientation.x\",\n",
    "        \"orientation.y\",\n",
    "        \"orientation.z\",\n",
    "        \"orientation.w\",\n",
    "        \"angVelocity.x\",\n",
    "        \"angVelocity.y\",\n",
    "        \"angVelocity.z\",\n",
    "        \"velocity.linear.x\",\n",
    "        \"velocity.linear.y\",\n",
    "        \"velocity.linear.z\",\n",
    "    ],\n",
    ")\n",
    "autopilotInfo_msg = Msg_structure(\n",
    "    name=name_space + \"AutopilotInfo\",\n",
    "    feature_list=[\n",
    "        \"Time\",\n",
    "        \"VelocityDesired.x\",\n",
    "        \"VelocityDesired.y\",\n",
    "        \"VelocityDesired.z\",\n",
    "    ],\n",
    "    feature_rename_list=[\n",
    "        \"Time\",\n",
    "        \"target.velocity.x\",\n",
    "        \"target.velocity.y\",\n",
    "        \"target.velocity.z\",\n",
    "    ],\n",
    ")\n",
    "\n",
    "if data_sim_not_real:\n",
    "    obs_msg = [Imu_msg, orientation_msg, autopilotInfo_msg]\n",
    "else:\n",
    "    obs_msg = [\n",
    "        orientation_msg,\n",
    "        Imu_msg,\n",
    "        autopilotInfo_msg,\n",
    "        transmitter_msg,\n",
    "    ]\n",
    "\n",
    "actuator_command_msg = Msg_structure(\n",
    "    name=\"actuatorcommand\",\n",
    "    feature_list=[\"Time\", \"data.data\"],\n",
    "    feature_rename_list=[\n",
    "        \"Time\",\n",
    "        \"m2\",\n",
    "        \"lfin\",\n",
    "        \"rfin\",\n",
    "        \"tfin\",\n",
    "        \"bfin\",\n",
    "        \"stick\",\n",
    "        \"m1\",\n",
    "        \"unused0\",\n",
    "        \"m0\",\n",
    "        \"unused1\",\n",
    "        \"unused2\",\n",
    "        \"unused3\",\n",
    "    ],\n",
    ")\n",
    "actuator_msg = Msg_structure(\n",
    "    name=name_space + \"actuators\",\n",
    "    feature_list=[\"Time\", \"data.data\"],\n",
    "    feature_rename_list=[\n",
    "        \"Time\",\n",
    "        \"m2\",\n",
    "        \"lfin\",\n",
    "        \"rfin\",\n",
    "        \"tfin\",\n",
    "        \"bfin\",\n",
    "        \"stick\",\n",
    "        \"m1\",\n",
    "        \"unused0\",\n",
    "        \"m0\",\n",
    "        \"unused1\",\n",
    "        \"unused2\",\n",
    "        \"unused3\",\n",
    "    ],\n",
    ")\n",
    "# act_msg = [actuator_msg, actuator_command_msg] #TODO: add act command as training data as well\n",
    "act_msg = [actuator_msg]\n",
    "all_msg = obs_msg + act_msg"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Try not to modify this"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load data\n",
    "bool_convert_rosbag2csv = True\n",
    "\n",
    "num_transitions = 2\n",
    "assert num_transitions >= 2, \"num_transitions should be greater than 2\"\n",
    "\n",
    "obs_feature_msg = [msg.feature_rename_list for msg in obs_msg]\n",
    "act_feature_msg = [msg.feature_rename_list for msg in act_msg]\n",
    "obs_feature_list = list(set([item for sublist in obs_feature_msg for item in sublist]))\n",
    "act_feature_list = list(set([item for sublist in act_feature_msg for item in sublist]))\n",
    "\n",
    "time_feature_name = \"Time\"\n",
    "simplify_control_list = [\"rfin\", \"bfin\", \"m1\", \"m2\"]\n",
    "purge_feature_list = [\n",
    "    \"ROSControlled\",\n",
    "    \"Armed\",\n",
    "    \"Thrust\",\n",
    "    \"unused0\",\n",
    "    \"unused1\",\n",
    "    \"unused2\",\n",
    "    \"unused3\",\n",
    "]\n",
    "purge_feature_list = purge_feature_list + simplify_control_list\n",
    "purge_feature_list.append(time_feature_name)\n",
    "\n",
    "merge_feature = [time_feature_name]\n",
    "merge_feature = merge_feature.append(act_feature_list)\n",
    "\n",
    "# Data processing\n",
    "round_time = 1\n",
    "assert round_time >= 1, \"round_time should be greater than 1\"\n",
    "\n",
    "bool_scale_obs = True\n",
    "obs_filter_feature = [\n",
    "    member\n",
    "    for sublist in obs_feature_msg\n",
    "    for member in sublist\n",
    "    if member not in purge_feature_list\n",
    "]\n",
    "obs_range_dic = {\n",
    "    \"linear_acceleration.x\": (-8, 18),\n",
    "    \"linear_acceleration.y\": (-5, 4),\n",
    "    \"linear_acceleration.z\": (8, 30),\n",
    "    \"orientation.x\": (-0.2, 0.2),\n",
    "    \"orientation.y\": (-0.3, 0.3),\n",
    "    \"orientation.z\": (-1, 1),\n",
    "    \"orientation.w\": (0, 1),\n",
    "    \"angVelocity.x\": (-25, 25),\n",
    "    \"angVelocity.y\": (-20, 60),\n",
    "    \"angVelocity.z\": (-30, 30),\n",
    "    \"velocity.linear.x\": (-10, 10),\n",
    "    \"velocity.linear.y\": (-10, 10),\n",
    "    \"velocity.linear.z\": (-4, 8),\n",
    "    \"target.velocity.x\": (-10, 10),\n",
    "    \"target.velocity.y\": (-10, 10),\n",
    "    \"target.velocity.z\": (-4, 8),\n",
    "}\n",
    "obs_scale_bnd = {\"scaled_obs_range\": (-1, 1)}\n",
    "for key in obs_filter_feature:\n",
    "    obs_scale_bnd[key] = obs_range_dic[key]\n",
    "\n",
    "bool_scale_act = True\n",
    "act_filter_feature = actuator_msg.feature_rename_list\n",
    "act_filter_feature = [\n",
    "    member for member in act_filter_feature if member not in purge_feature_list\n",
    "]\n",
    "act_scale_bnd = {\n",
    "    \"act_range\": (1000, 2000),\n",
    "    \"scaled_act_range\": (-1, 1),\n",
    "}\n",
    "\n",
    "bool_addTarget = True\n",
    "num_target_df = 7\n",
    "pseudo_target_range_list = [(-7, 7), (-7, 7), (-1, 1)]\n",
    "pseudo_target_freq_range_list = [(0.05, 0.001), (0.05, 0.001), (0.1, 0.001)]\n",
    "target_feature_list = [\"target.velocity.x\", \"target.velocity.y\", \"target.velocity.z\"]\n",
    "pseudo_target_dict = {\n",
    "    \"num_target\": num_target_df,\n",
    "    \"target_feature_list\": target_feature_list,\n",
    "    \"range_list\": pseudo_target_range_list,\n",
    "    \"freq_range_list\": pseudo_target_freq_range_list,\n",
    "}\n",
    "\n",
    "bool_addNoise = True\n",
    "num_noise_df = 3\n",
    "noise_level = 0.05\n",
    "\n",
    "bool_merge_target_vel = True\n",
    "target_feature_list = target_feature_list\n",
    "vel_feature_list = [\"velocity.linear.x\", \"velocity.linear.y\", \"velocity.linear.z\"]\n",
    "diff_vel_name = [\"diff_vel.x\", \"diff_vel.y\", \"diff_vel.z\"]\n",
    "diff_vel_filter_feature = obs_filter_feature + diff_vel_name\n",
    "diff_vel_filter_feature = [\n",
    "    member\n",
    "    for member in diff_vel_filter_feature\n",
    "    if member not in (target_feature_list + vel_feature_list)\n",
    "]\n",
    "\n",
    "train_test_split_ratio = (0.99, 0.01)\n",
    "\n",
    "# Reward Function\n",
    "xy_target_feature_list = target_feature_list[0:1]\n",
    "xy_cost_feature_list = vel_feature_list[0:1]\n",
    "z_target_feature_list = target_feature_list[2]\n",
    "z_cost_feature_list = vel_feature_list[2]\n",
    "acc_feature_list = [\n",
    "    \"linear_acceleration.x\",\n",
    "    \"linear_acceleration.y\",\n",
    "    \"linear_acceleration.z\",\n",
    "]\n",
    "actCost_feature_list = [\"m0\", \"m1\", \"m2\"]\n",
    "actCost_feature_list = [\n",
    "    member for member in actCost_feature_list if member not in purge_feature_list\n",
    "]\n",
    "actChangeCost_feature_list = [\"m0\", \"m1\", \"m2\", \"stick\", \"tfin\", \"bfin\", \"lfin\", \"rfin\"]\n",
    "actChangeCost_feature_list = [\n",
    "    member for member in actChangeCost_feature_list if member not in purge_feature_list\n",
    "]\n",
    "\n",
    "weight_next_obs = 1\n",
    "weight_prev_obs = 0 * 1 / (num_transitions - 1)\n",
    "weight_prev_act = 0.4 * 1 / (num_transitions - 1)\n",
    "weight_prev_actChange = 0 * 1 / (num_transitions - 1)\n",
    "reward_weights = (\n",
    "    weight_next_obs,\n",
    "    weight_prev_obs,\n",
    "    weight_prev_act,\n",
    "    weight_prev_actChange,\n",
    ")\n",
    "\n",
    "if policy_xy_not_z:\n",
    "    weight_xy_obs, weigth_z_obs = 1, 0\n",
    "else:\n",
    "    weight_xy_obs, weigth_z_obs = 0, 1\n",
    "weight_acc_obs = 0.4\n",
    "obs_reward_weights = (weight_xy_obs, weigth_z_obs, weight_acc_obs)\n",
    "acceleration_constant = -0.83\n",
    "\n",
    "params = {\n",
    "    \"exp_index\": exp_index,\n",
    "    \"data_folder_path\": data_folder_path,\n",
    "    \"RealResults_path\": RealResults_path,\n",
    "    \"SimResults_path\": SimResults_path,\n",
    "    \"MixResults_path\": MixResults_path,\n",
    "    \"bool_convert_rosbag2csv\": bool_convert_rosbag2csv,\n",
    "    \"all_msg\": all_msg,\n",
    "    \"obs_msg\": obs_msg,\n",
    "    \"act_msg\": act_msg,\n",
    "    \"num_transitions\": num_transitions,\n",
    "    \"target_feature_list\": target_feature_list,\n",
    "    \"xy_target_feature_list\": xy_target_feature_list,\n",
    "    \"z_target_feature_list\": z_target_feature_list,\n",
    "    \"obs_feature_list\": obs_feature_list,\n",
    "    \"act_feature_list\": act_feature_list,\n",
    "    \"time_feature_name\": time_feature_name,\n",
    "    \"purge_feature_list\": purge_feature_list,\n",
    "    \"round_time\": round_time,\n",
    "    \"act_filter_feature\": act_filter_feature,\n",
    "    \"obs_filter_feature\": obs_filter_feature,\n",
    "    \"train_test_split_ratio\": train_test_split_ratio,\n",
    "    \"bool_scale_obs\": bool_scale_obs,\n",
    "    \"bool_scale_act\": bool_scale_act,\n",
    "    \"act_scale_bnd\": act_scale_bnd,\n",
    "    \"obs_scale_bnd\": obs_scale_bnd,\n",
    "    \"bool_addTarget\": bool_addTarget,\n",
    "    \"pseudo_target_dict\": pseudo_target_dict,\n",
    "    \"bool_addNoise\": bool_addNoise,\n",
    "    \"num_noise_df\": num_noise_df,\n",
    "    \"noise_level\": noise_level,\n",
    "    \"bool_merge_target_vel\": bool_merge_target_vel,\n",
    "    \"diff_vel_filter_feature\": diff_vel_filter_feature,\n",
    "    \"vel_feature_list\": vel_feature_list,\n",
    "    \"xy_cost_feature_list\": xy_cost_feature_list,\n",
    "    \"z_cost_feature_list\": z_cost_feature_list,\n",
    "    \"actCost_feature_list\": actCost_feature_list,\n",
    "    \"actChangeCost_feature_list\": actChangeCost_feature_list,\n",
    "    \"reward_weights\": reward_weights,\n",
    "    \"obs_reward_weights\": obs_reward_weights,\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Convert rosbag to csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data_folder_path: ./blimpSim/data/behaviour_policy\n",
      "----------Start Converting Rosbag to CSV files----------\n",
      "bag: square.bag already converted\n",
      "bag: lines.bag already converted\n",
      "bag: reverse_lines.bag already converted\n",
      "bag: reverse_coil.bag already converted\n",
      "bag: reverse_down_coil.bag already converted\n",
      "bag: double_square.bag already converted\n",
      "bag: coil.bag already converted\n",
      "bag: lines2.bag already converted\n",
      "bag: manual.bag already converted\n",
      "bag: reverse_lines2.bag already converted\n",
      "bag: double_coil.bag already converted\n",
      "bag: reverse_double_coil.bag already converted\n",
      "bag: down_coil.bag already converted\n",
      "----------Finish Convert----------\n"
     ]
    }
   ],
   "source": [
    "if params[\"bool_convert_rosbag2csv\"]:\n",
    "    import bagpy\n",
    "    from bagpy import bagreader\n",
    "    from pathlib import Path\n",
    "\n",
    "    class ConvertBag2csv:\n",
    "        def rosbag2csv(self, rosbag_path):\n",
    "            b = bagreader(rosbag_path)\n",
    "            csvfiles = []\n",
    "            for t in b.topics:\n",
    "                data = b.message_by_topic(t)\n",
    "                csvfiles.append(data)\n",
    "\n",
    "        def convert_rosbag2csv(self, directory):\n",
    "            for root, dirs, files in os.walk(directory):\n",
    "                for file in files:\n",
    "                    if file.endswith(\".bag\"):\n",
    "                        try:\n",
    "                            bag_path = os.path.join(root, file)\n",
    "                            file_path = os.path.splitext(bag_path)[0]\n",
    "                            if not os.path.exists(file_path):\n",
    "                                print(f\"bag: start convert {file} ...\")\n",
    "                                self.rosbag2csv(bag_path)\n",
    "                            else:\n",
    "                                print(f\"bag: {file} already converted\")\n",
    "                        except ValueError as e:\n",
    "                            print(f\"bag: not able to convert {file}\")\n",
    "                            raise ConvertBag2csv(\"hold brocken bag file\") from e\n",
    "\n",
    "    print(f\"data_folder_path: {data_folder_path}\")\n",
    "    clsConvertBag2csv = ConvertBag2csv()\n",
    "\n",
    "    print(\"----------Start Converting Rosbag to CSV files----------\")\n",
    "    clsConvertBag2csv.convert_rosbag2csv(data_folder_path)\n",
    "    print(\"----------Finish Convert----------\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define Transition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "class objTransitions:\n",
    "    def __init__(\n",
    "        self,\n",
    "        num_transitions=2,\n",
    "        act_filter_feature=[],\n",
    "        obs_filter_feature=[],\n",
    "        purge_feature_list=[\"Time\"],\n",
    "        time_feature_name=[\"Time\"],\n",
    "        **kwargs\n",
    "    ):\n",
    "        self._df = None\n",
    "        self.num_transitions = num_transitions\n",
    "        assert self.num_transitions >= 2\n",
    "\n",
    "        self.act_filter_feature = act_filter_feature\n",
    "        self.obs_filter_feature = obs_filter_feature\n",
    "        self.purge_feature_list = purge_feature_list\n",
    "        self.time_feature_name = time_feature_name\n",
    "\n",
    "        self.time = None\n",
    "        self.act_trans_list = []\n",
    "        self.obs_trans_list = []\n",
    "\n",
    "    @property\n",
    "    def df(self):\n",
    "        return self._df\n",
    "\n",
    "    @df.setter\n",
    "    def df(self, df):\n",
    "        self._df = df\n",
    "        if self._df is not None:\n",
    "            self._update()\n",
    "\n",
    "    def _update(self):\n",
    "        self.df_len = len(self._df.index)\n",
    "        self.trans_len = len(self._df.index) - self.num_transitions\n",
    "        self.time = self.get_time()\n",
    "        self.act_trans_list, self.obs_trans_list = self._split_df(self._df)\n",
    "        self._reset_index()\n",
    "\n",
    "    def _calculate_delta_obs(self, obs, next_obs):\n",
    "        return next_obs - obs\n",
    "\n",
    "    def _split_df(self, df):\n",
    "        act_trans_list, obs_trans_list = [], []\n",
    "        for i in range(self.num_transitions):\n",
    "            act_obs_trans = df[i : self.df_len - self.num_transitions + 1 + i]\n",
    "            act_obs_trans.reset_index(drop=True)\n",
    "            act_trans, obs_trans = (\n",
    "                act_obs_trans[self.act_filter_feature],\n",
    "                act_obs_trans[self.obs_filter_feature],\n",
    "            )\n",
    "            act_trans, obs_trans = self._purge_feature(act_trans), self._purge_feature(\n",
    "                obs_trans\n",
    "            )\n",
    "            act_trans_list.append(act_trans)\n",
    "            obs_trans_list.append(obs_trans)\n",
    "        return act_trans_list, obs_trans_list\n",
    "\n",
    "    def _purge_feature(self, df):\n",
    "        for feature in self.purge_feature_list:\n",
    "            if feature in df.columns:\n",
    "                print(\"[ objTransition ] purge feature from df: \", feature)\n",
    "                df = df.drop(feature, axis=1)\n",
    "        return df\n",
    "\n",
    "    def _reset_index(self):\n",
    "        self._df.reset_index(drop=True, inplace=True)\n",
    "        for i in range(self.num_transitions):\n",
    "            self.act_trans_list[i].reset_index(drop=True, inplace=True)\n",
    "            self.obs_trans_list[i].reset_index(drop=True, inplace=True)\n",
    "\n",
    "    def get_next_transitions(self):\n",
    "        next_act, next_obs = (\n",
    "            self.act_trans_list[self.num_transitions - 1],\n",
    "            self.obs_trans_list[self.num_transitions - 1],\n",
    "        )\n",
    "        return next_act, next_obs\n",
    "\n",
    "    def get_cur_transitions(self):\n",
    "        act, obs = (\n",
    "            self.act_trans_list[self.num_transitions - 2],\n",
    "            self.obs_trans_list[self.num_transitions - 2],\n",
    "        )\n",
    "        return act, obs\n",
    "\n",
    "    def get_prev_transitions_list(self):\n",
    "        prev_act_df_list, prev_obs_df_list = (\n",
    "            self.act_trans_list[: self.num_transitions - 1],\n",
    "            self.obs_trans_list[: self.num_transitions - 1],\n",
    "        )\n",
    "        return prev_act_df_list, prev_obs_df_list\n",
    "\n",
    "    def get_delta_obs_transitions(self):\n",
    "        _, next_obs = self.get_next_transitions()\n",
    "        _, cur_obs = self.get_cur_transitions()\n",
    "        delta_obs = self._calculate_delta_obs(cur_obs, next_obs)\n",
    "        return delta_obs\n",
    "\n",
    "    def get_time(self):\n",
    "        return self._df[self.time_feature_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Cleaner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Data_cleaner:\n",
    "    def clean_df(self, df, act_filter_feature, act_scale_bnd, round_time, **params):\n",
    "        df = self.fill_df_NA(df)\n",
    "        df = self.clean_act_bnd_error(df, act_scale_bnd, act_filter_feature)\n",
    "        df = self.clean_time_sequence(df, round_time)\n",
    "        return df\n",
    "\n",
    "    def fill_df_NA(self, df):\n",
    "        df = df.fillna(method=\"ffill\")\n",
    "        df = df.fillna(0)\n",
    "        return df\n",
    "\n",
    "    def clean_act_bnd_error(self, df, act_scale_bnd, act_filter_feature):\n",
    "        act_min, act_max = act_scale_bnd[\"act_range\"]\n",
    "        df = self.clip_df(df, act_min, act_max, act_filter_feature)\n",
    "        return df\n",
    "\n",
    "    def clip_df(self, df, df_min, df_max, df_filter_feature):\n",
    "        for filter_feature in df_filter_feature:\n",
    "            df = df[df[filter_feature] >= df_min]\n",
    "            df = df[df[filter_feature] <= df_max]\n",
    "        df.reset_index(drop=True, inplace=True)\n",
    "        return df\n",
    "\n",
    "    def clean_time_sequence(self, df, round_time=1):\n",
    "        time = df[\"Time\"]\n",
    "        time = time.round(round_time)\n",
    "        time = time.drop_duplicates()\n",
    "        df = df.loc[time.index]\n",
    "        return df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Loader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "class Data_loader:\n",
    "    def load_csv(self, directory, all_msg):\n",
    "        csv_index = self.load_csv_index(directory)\n",
    "        self.check_all_required_csv(csv_index, all_msg)\n",
    "        return csv_index\n",
    "\n",
    "    def load_csv_index(self, directory):\n",
    "        csv_index = []\n",
    "        for root, dirs, files in os.walk(directory):\n",
    "            for file in files:\n",
    "                if file.endswith(\".csv\"):\n",
    "                    path = os.path.join(root, file)\n",
    "                    csv_index.append(path)\n",
    "        return csv_index\n",
    "\n",
    "    def check_all_required_csv(self, csv_index, all_msg):\n",
    "        for msg in all_msg:\n",
    "            csv_not_exist = self.csv_not_exist(csv_index, msg.name)\n",
    "            if csv_not_exist:\n",
    "                print(f\"msg: {msg.name} is required but csv file does not exist\")\n",
    "                raise\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def create_df_from_csvIndex(self, csv_index, **params):\n",
    "        df_list = self.read_csvfiles(csv_index, **params)\n",
    "        df = self.merge_df_list(df_list, merge_feature)\n",
    "        df = df.sort_values(by=\"Time\", ignore_index=True)\n",
    "        return df\n",
    "\n",
    "    def read_csvfiles(self, csv_index, **params):\n",
    "        df_list = []\n",
    "        for msg in all_msg:\n",
    "            print(\"[ Load Data ] read_csvfiles: \", msg.name)\n",
    "\n",
    "            if (msg.name == \"actuatorcommand\") and self.csv_not_exist(\n",
    "                csv_index, msg.name\n",
    "            ):\n",
    "                print(\n",
    "                    \"[ Load Data ] read_csvfiles, actuatorcommand.csv does not exist, skip...\"\n",
    "                )\n",
    "\n",
    "            else:\n",
    "                df = self.create_df(\n",
    "                    csv_index,\n",
    "                    msg.name,\n",
    "                    msg.feature_list,\n",
    "                    msg.feature_rename_list,\n",
    "                    **params,\n",
    "                )\n",
    "                df_list.append(df)\n",
    "        return df_list\n",
    "\n",
    "    def create_df(\n",
    "        self, csv_index, msg_name, feature_list, feature_rename_list, **params\n",
    "    ):\n",
    "        df = self.read_df_from_msg_in_csvIndex(csv_index, msg_name, **params)\n",
    "        df = df[feature_list]\n",
    "        df = self.unwrap_features(df)\n",
    "        df.columns = feature_rename_list\n",
    "        return df\n",
    "\n",
    "    def read_df_from_msg_in_csvIndex(self, csv_index, msg_name, **params):\n",
    "        for csv_name in csv_index:\n",
    "            if msg_name in csv_name:\n",
    "                print(csv_name)\n",
    "                df = pd.read_csv(csv_name)\n",
    "        return df\n",
    "\n",
    "    def merge_df_list(self, df_list, merge_feature):\n",
    "        df = df_list[0]\n",
    "        iter_data = iter(df_list)\n",
    "        next(iter_data)\n",
    "        for item in iter_data:\n",
    "            df = pd.merge(df, item, how=\"outer\", on=merge_feature)\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def unwrap_features(self, df):\n",
    "        for col_name in df.columns:\n",
    "            col = df[col_name]\n",
    "            col_data_type = self.check_col_data_type(col)\n",
    "\n",
    "            if col_data_type == tuple:\n",
    "                df = self.unwrap_tuple_feature(df, col_name)\n",
    "\n",
    "        return df\n",
    "\n",
    "    def check_col_data_type(self, col):\n",
    "        first_element = col[0]\n",
    "        data_type = type(first_element)\n",
    "\n",
    "        if data_type == str:\n",
    "            col = col.map(lambda x: eval(x))\n",
    "            first_element = col[0]\n",
    "            data_type = type(first_element)\n",
    "\n",
    "        return data_type\n",
    "\n",
    "    def unwrap_tuple_feature(self, df, col_name):\n",
    "        tuple_col = df.pop(col_name)\n",
    "        tuple_col = tuple_col.map(lambda x: eval(x))\n",
    "\n",
    "        first_element = tuple_col[0]\n",
    "        feature_len = len(first_element)\n",
    "\n",
    "        feature_list = []\n",
    "        for i in range(feature_len):\n",
    "            new_feature = tuple_col.str[i]\n",
    "            feature_list.append(new_feature)\n",
    "\n",
    "        feature_pd = pd.concat(feature_list, axis=1)\n",
    "        df = pd.concat([df, feature_pd], axis=1)\n",
    "\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def csv_not_exist(self, csv_index, msg_name):\n",
    "        target_csv = [csv_name for csv_name in csv_index if msg_name in csv_name]\n",
    "        if len(target_csv) == 0:\n",
    "            return True\n",
    "        else:\n",
    "            return False\n",
    "\n",
    "    def concat_df(self, df, df_new):\n",
    "        if df is None:\n",
    "            return df_new\n",
    "        else:\n",
    "            df = pd.concat([df, df_new], axis=0, ignore_index=True)\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def main(self, **params):\n",
    "        df = None\n",
    "\n",
    "        for exp in exp_index:\n",
    "            exp_path = os.path.join(data_folder_path, exp)\n",
    "\n",
    "            print(\"[ LOAD csv ]\", exp)\n",
    "            csv_index = self.load_csv(exp_path, all_msg)\n",
    "\n",
    "            print(\"[ CREATE df ] start\", csv_index)\n",
    "            new_df = self.create_df_from_csvIndex(csv_index, **params)\n",
    "\n",
    "            print(\"[ Clean df ] start\")\n",
    "            clsDataCleaner = Data_cleaner()\n",
    "            new_df = clsDataCleaner.clean_df(new_df, **params)\n",
    "\n",
    "            print(\"[ CONCAT df ] start\")\n",
    "            df = self.concat_df(df, new_df)\n",
    "\n",
    "        print(\"[ Create Transitions ] Start\")\n",
    "        objTrans = objTransitions(**params)\n",
    "        objTrans.df = df\n",
    "\n",
    "        print(\"[ LOAD csv ] Complete\")\n",
    "        return objTrans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ LOAD csv ] coil\n",
      "[ CREATE df ] start ['./blimpSim/data/behaviour_policy/coil/blimp-tail-position.csv', './blimpSim/data/behaviour_policy/coil/blimp-TransmitterInfo.csv', './blimpSim/data/behaviour_policy/coil/blimp-ground_truth-imu.csv', './blimpSim/data/behaviour_policy/coil/blimp-tail-pose.csv', './blimpSim/data/behaviour_policy/coil/blimp-actuators.csv', './blimpSim/data/behaviour_policy/coil/blimp-GCSACTUATORS.csv', './blimpSim/data/behaviour_policy/coil/blimp-pose.csv', './blimpSim/data/behaviour_policy/coil/blimp-AutopilotInfo.csv', './blimpSim/data/behaviour_policy/coil/blimp-ground_speed.csv', './blimpSim/data/behaviour_policy/coil/blimp-tail-imu.csv', './blimpSim/data/behaviour_policy/coil/blimp-ground_truth-pose.csv', './blimpSim/data/behaviour_policy/coil/blimp-tail-transform.csv', './blimpSim/data/behaviour_policy/coil/blimp-ground_truth-position.csv']\n",
      "[ Load Data ] read_csvfiles:  blimp-tail-imu\n",
      "./blimpSim/data/behaviour_policy/coil/blimp-tail-imu.csv\n",
      "[ Load Data ] read_csvfiles:  blimp-pose\n",
      "./blimpSim/data/behaviour_policy/coil/blimp-pose.csv\n",
      "[ Load Data ] read_csvfiles:  blimp-AutopilotInfo\n",
      "./blimpSim/data/behaviour_policy/coil/blimp-AutopilotInfo.csv\n",
      "[ Load Data ] read_csvfiles:  blimp-actuators\n",
      "./blimpSim/data/behaviour_policy/coil/blimp-actuators.csv\n",
      "[ Clean df ] start\n",
      "[ CONCAT df ] start\n",
      "[ LOAD csv ] double_coil\n",
      "[ CREATE df ] start ['./blimpSim/data/behaviour_policy/double_coil/blimp-tail-position.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-TransmitterInfo.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-ground_truth-imu.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-tail-pose.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-actuators.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-GCSACTUATORS.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-pose.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-AutopilotInfo.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-ground_speed.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-tail-imu.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-ground_truth-pose.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-tail-transform.csv', './blimpSim/data/behaviour_policy/double_coil/blimp-ground_truth-position.csv']\n",
      "[ Load Data ] read_csvfiles:  blimp-tail-imu\n",
      "./blimpSim/data/behaviour_policy/double_coil/blimp-tail-imu.csv\n",
      "[ Load Data ] read_csvfiles:  blimp-pose\n",
      "./blimpSim/data/behaviour_policy/double_coil/blimp-pose.csv\n",
      "[ Load Data ] read_csvfiles:  blimp-AutopilotInfo\n",
      "./blimpSim/data/behaviour_policy/double_coil/blimp-AutopilotInfo.csv\n",
      "[ Load Data ] read_csvfiles:  blimp-actuators\n",
      "./blimpSim/data/behaviour_policy/double_coil/blimp-actuators.csv\n",
      "[ Clean df ] start\n",
      "[ CONCAT df ] start\n",
      "[ Create Transitions ] Start\n",
      "[ LOAD csv ] Complete\n"
     ]
    }
   ],
   "source": [
    "clsLoadData = Data_loader()\n",
    "objTrans = clsLoadData.main(**params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fe4734c1e80>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe47369b610>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe477a77160>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe477b1d9d0>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cur_act, cur_obs = objTrans.get_cur_transitions()\n",
    "cur_act.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fe47346a190>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe473708370>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe47371fe50>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe47348c1c0>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe473776490>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe47374a820>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe47374a910>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe47368edc0>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe473646640>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe4734eba90>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe4733a2ee0>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe4735c9370>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe4737b37c0>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe4737bec10>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe4737fdd30>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe4733504f0>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 16 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cur_obs.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        40.720\n",
       "1        40.750\n",
       "2        40.855\n",
       "3        40.950\n",
       "4        41.055\n",
       "         ...   \n",
       "4512    530.350\n",
       "4513    530.455\n",
       "4514    530.550\n",
       "4515    530.655\n",
       "4516    530.750\n",
       "Name: Time, Length: 4517, dtype: float64"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "objTrans.get_time()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data processing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Real Data Filter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "class RealDataFilter:\n",
    "    def compute_flight_df(self, df):\n",
    "        time, armed, thrust = df[\"Time\"], df[\"Armed\"], df[\"Thrust\"]\n",
    "\n",
    "        arm_start_time, arm_stop_time = self.get_arm_start_end_time(armed, time)\n",
    "        flight_start_time, flight_stop_time = self.get_flight_time(\n",
    "            time, thrust, arm_start_time, arm_stop_time\n",
    "        )\n",
    "        flight_df = self.get_flight_df(df, time, flight_start_time, flight_stop_time)\n",
    "\n",
    "        return flight_df\n",
    "\n",
    "    def get_arm_start_end_time(self, armed, time):\n",
    "        armedValue = armed[armed.isna() != 1]\n",
    "        tArmedValue = time[armed.isna() != 1]\n",
    "\n",
    "        ta_np = tArmedValue.to_numpy()\n",
    "        av_np = armedValue.to_numpy()\n",
    "\n",
    "        gav_np = np.gradient(av_np)\n",
    "        idxStart = np.where(gav_np > 0)\n",
    "        idxStop = np.where(gav_np < 0)\n",
    "\n",
    "        tArmedStart = ta_np[idxStart]\n",
    "        tArmedStop = ta_np[idxStop]\n",
    "\n",
    "        tArmedStart = tArmedStart[range(0, len(tArmedStart), 2)]\n",
    "        tArmedStop = tArmedStop[range(0, len(tArmedStop), 2)]\n",
    "\n",
    "        if tArmedStart.shape[0] > tArmedStop.shape[0]:\n",
    "            tArmedStop = np.append(tArmedStop, ta_np[-1])\n",
    "\n",
    "        return tArmedStart, tArmedStop\n",
    "\n",
    "    def get_flight_time(\n",
    "        self,\n",
    "        time,\n",
    "        thrust,\n",
    "        tArmedStart,\n",
    "        tArmedStop,\n",
    "        thrustThreshold=0.15,\n",
    "        tDeltaThrust=10,\n",
    "        tDeltaDisarm=15,\n",
    "    ):\n",
    "        \"\"\"\n",
    "        - thrust_thresh: Thrust threshold. Flight is considered to be started when vehicel is\n",
    "                    armed and for the first time the thrust is higher than threshold value\n",
    "        - time_deltaThrust_thresh: Time delta added to time point where vehicle is armed and\n",
    "                    thrust value crosses threshold. During take off vehicle usually is pushed\n",
    "                    around by helpers, that is why this time delta is necessary to exclude the\n",
    "                    take off phase from the data\n",
    "        - time_deltaDisarm: Since vehicle is disarmed in advance of a crash or a landing or at\n",
    "                    least shortly after, tDeltaDisarm  is subtracted from the time point when\n",
    "                    disarming takes place. That way it is made sure that only flight data is considered\n",
    "        \"\"\"\n",
    "\n",
    "        tFlightStart, tFlightStop = [], []\n",
    "        for i in range(len(tArmedStart)):\n",
    "            tArmedStartTemp = tArmedStart[i]\n",
    "            tArmedStopTemp = tArmedStop[i]\n",
    "\n",
    "            thrustTemp = thrust[(time > tArmedStartTemp) & (time < tArmedStopTemp)]\n",
    "            tThrustTemp = time[(time > tArmedStartTemp) & (time < tArmedStopTemp)]\n",
    "\n",
    "            if thrustTemp[thrustTemp > thrustThreshold].empty:\n",
    "                continue\n",
    "\n",
    "            idxThrustThreshold = thrustTemp[thrustTemp > thrustThreshold].index[0]\n",
    "            tFlightStartTemp = tThrustTemp[idxThrustThreshold] + tDeltaThrust\n",
    "            tFlightStopTemp = tArmedStopTemp - tDeltaDisarm\n",
    "\n",
    "            tFlightStart.append(tFlightStartTemp)\n",
    "            tFlightStop.append(tFlightStopTemp)\n",
    "\n",
    "        return tFlightStart, tFlightStop\n",
    "\n",
    "    def get_flight_df(self, df, time, tFlightStart, tFlightStop):\n",
    "        dfFlightList = []\n",
    "        dfFlight = []\n",
    "        for i in range(len(tFlightStart)):\n",
    "            dfTemp = df[(time > tFlightStart[i]) & (time < tFlightStop[i])]\n",
    "            dfFlightList.append(dfTemp)\n",
    "            dfFlight = pd.concat(dfFlightList, ignore_index=True)\n",
    "        return dfFlight\n",
    "\n",
    "    def main(self, df, **params):\n",
    "        df = self.compute_flight_df(df)\n",
    "        return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "if not data_sim_not_real:\n",
    "    print(\" Filter Flight Time data\")\n",
    "    clsRealDataFilter = RealDataFilter()\n",
    "    objTrans.df = clsRealDataFilter.main(objTrans.df, **params)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Processor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Data_processor:\n",
    "    def data_processsing(self, df, **params):\n",
    "        if bool_addTarget:\n",
    "            df = self.duplicate_target_df(df, **pseudo_target_dict)\n",
    "        if bool_addNoise:\n",
    "            df = self.duplicate_Noise_df(df, num_noise_df, noise_level)\n",
    "        if bool_scale_obs:\n",
    "            df = self.scale_df_obs(df, obs_scale_bnd, obs_filter_feature)\n",
    "        if bool_scale_act:\n",
    "            df = self.scale_df_act(df, act_scale_bnd, act_filter_feature)\n",
    "            df = self.clip_scaled_df(df, act_scale_bnd, act_filter_feature)\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def scale_df_act(self, df, act_scale_bnd, act_filter_feature):\n",
    "        new_df = copy.deepcopy(df)\n",
    "        act = new_df[act_filter_feature]\n",
    "\n",
    "        act_min, act_max = act_scale_bnd[\"act_range\"]\n",
    "        scaled_act_min, scaled_act_max = act_scale_bnd[\"scaled_act_range\"]\n",
    "\n",
    "        act = self.minmax_scale_df(\n",
    "            act, act_max, act_min, scaled_act_max, scaled_act_min\n",
    "        )\n",
    "        new_df[act_filter_feature] = act\n",
    "        return new_df\n",
    "\n",
    "    def scale_df_obs(self, df, obs_scale_bnd, obs_filter_feature):\n",
    "        new_df = copy.deepcopy(df)\n",
    "        scaled_obs_min, scaled_obs_max = obs_scale_bnd[\"scaled_obs_range\"]\n",
    "        for feature in obs_filter_feature:\n",
    "            feature_min, feature_max = obs_scale_bnd[feature]\n",
    "            feature_df = new_df[feature]\n",
    "            feature_df = self.minmax_scale_df(\n",
    "                feature_df, feature_max, feature_min, scaled_obs_max, scaled_obs_min\n",
    "            )\n",
    "            new_df[feature] = feature_df\n",
    "        return new_df\n",
    "\n",
    "    def minmax_scale_df(self, df, df_max, df_min, target_max, target_min):\n",
    "        df_mean = (df_max + df_min) / 2\n",
    "        target_mean = (target_max + target_min) / 2\n",
    "\n",
    "        scale = (df_max - df_min) / (target_max - target_min)\n",
    "        df = (df - df_mean) / scale + target_mean\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def duplicate_target_df(\n",
    "        self, df, num_target, range_list, freq_range_list, target_feature_list, seed=None, **kwargs\n",
    "    ):\n",
    "        if seed is not None:\n",
    "            np.random.seed(seed)\n",
    "        time_df = df[\"Time\"]\n",
    "        df_list = []\n",
    "        df_list.append(df)\n",
    "        for i in range(num_target):\n",
    "            new_df = copy.deepcopy(df)\n",
    "            pseudo_target_df = self.generate_pseudo_target(\n",
    "                time_df, range_list, freq_range_list\n",
    "            )\n",
    "            pseudo_target_df.columns = target_feature_list\n",
    "            new_df[target_feature_list] = pseudo_target_df\n",
    "            df_list.append(new_df)\n",
    "        return pd.concat(df_list, axis=0, ignore_index=True)\n",
    "\n",
    "    def generate_pseudo_target(\n",
    "        self, time_df, pseudo_target_range_list, pseudo_target_freq_range_list\n",
    "    ):\n",
    "        target_df_list = []\n",
    "        for i in range(len(pseudo_target_range_list)):\n",
    "            new_target_df = self.generate_pseudo_sinusoidal_signal(\n",
    "                time_df, pseudo_target_range_list[i], pseudo_target_freq_range_list[i]\n",
    "            )\n",
    "            target_df_list.append(new_target_df)\n",
    "        return pd.concat(target_df_list, axis=1)\n",
    "\n",
    "    def generate_pseudo_sinusoidal_signal(\n",
    "        self, time_df, pseudo_target_range, pseudo_target_freq_range\n",
    "    ):\n",
    "        time_np = time_df.to_numpy()\n",
    "        range_min, range_max = pseudo_target_range\n",
    "        freq_range_min, freq_range_max = pseudo_target_freq_range\n",
    "\n",
    "        range_sin, range_cos = (\n",
    "            np.random.uniform(range_min, range_max),\n",
    "            np.random.uniform(range_min, range_max),\n",
    "        )\n",
    "        freq_sin, freq_cos = (\n",
    "            np.random.uniform(freq_range_min, freq_range_max),\n",
    "            np.random.uniform(freq_range_min, freq_range_max),\n",
    "        )\n",
    "        const_sin, const_cos = np.random.uniform(-np.pi, np.pi), np.random.uniform(\n",
    "            -np.pi, np.pi\n",
    "        )\n",
    "\n",
    "        sinusoidal_signal_np = range_sin * np.sin(\n",
    "            freq_sin * (time_np + const_sin)\n",
    "        ) + range_cos * np.cos(freq_cos * (time_np + const_cos))\n",
    "\n",
    "        sinusoidal_signal_df = pd.DataFrame(sinusoidal_signal_np)\n",
    "        return sinusoidal_signal_df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def duplicate_Noise_df(self, df, num_noise_df, noise_level, seed=None):\n",
    "        if seed is not None:\n",
    "            np.random.seed(seed)\n",
    "        df_list = []\n",
    "        df_list.append(df)\n",
    "\n",
    "        time_df = df[\"Time\"]\n",
    "        new_df = copy.deepcopy(df)\n",
    "        new_df = new_df.drop(\"Time\", axis=1)\n",
    "        for i in range(num_noise_df):\n",
    "            noised_df = self.addNoise_df(new_df, noise_level)\n",
    "            noised_df = pd.concat([time_df, noised_df], axis=1)\n",
    "            df_list.append(noised_df)\n",
    "\n",
    "        return pd.concat(df_list, axis=0, ignore_index=True)\n",
    "\n",
    "    def addNoise_df(self, df, noise_level):\n",
    "        df_range = df.max() - df.min()\n",
    "        df_noise_range = noise_level * df_range / 2\n",
    "        noise = np.random.uniform(-df_noise_range, df_noise_range, df.shape)\n",
    "        df = df + noise\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def clip_scaled_df(self, df, act_scale_bnd, act_filter_feature):\n",
    "        new_df = copy.deepcopy(df)\n",
    "        scaled_act_min, scaled_act_max = act_scale_bnd[\"scaled_act_range\"]\n",
    "        new_df = self.clip_df(\n",
    "            new_df, scaled_act_min, scaled_act_max, act_filter_feature\n",
    "        )\n",
    "        return new_df\n",
    "\n",
    "    def clip_df(self, df, df_min, df_max, df_filter_feature):\n",
    "        for filter_feature in df_filter_feature:\n",
    "            df = df[df[filter_feature] >= df_min]\n",
    "            df = df[df[filter_feature] <= df_max]\n",
    "        df.reset_index(drop=True, inplace=True)\n",
    "        return df\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def split_trainTest_df(self, df, train_test_split_ratio, **params):\n",
    "        df_len = len(df.index)\n",
    "        idx = self.get_trainTest_index(df_len, train_test_split_ratio)\n",
    "        df_train, df_test = self.split_df(df, idx)\n",
    "        return df_train, df_test\n",
    "\n",
    "    def split_df(self, df, idx):\n",
    "        idx_train, idx_test = idx\n",
    "        df_train, df_test = df.loc[idx_train], df.loc[idx_test]\n",
    "        df_train = df_train.reset_index(drop=True)\n",
    "        df_test = df_test.reset_index(drop=True)\n",
    "        return df_train, df_test\n",
    "\n",
    "    def get_trainTest_index(self, n_data, ratio):\n",
    "        ratio_train, ratio_test = ratio\n",
    "        n_train = int(n_data * ratio_train)\n",
    "\n",
    "        idx_randperm = np.random.permutation(n_data - 1)\n",
    "        idx_train = np.sort(idx_randperm[:n_train])\n",
    "        idx_test = np.sort(idx_randperm[n_train + 1 :])\n",
    "        return (idx_train, idx_test)\n",
    "\n",
    "    # *******************************************\n",
    "\n",
    "    def main(self, df, **params):\n",
    "        print(\n",
    "            \"[ Data processing ] Scale action data, Add noise, Round data, clip data \"\n",
    "        )\n",
    "        df = self.data_processsing(df, **params)\n",
    "\n",
    "        print(\"[ Data processing ] Split train and test data \")\n",
    "        df_train, df_test = self.split_trainTest_df(df, **params)\n",
    "\n",
    "        print(\"[ Data processing ] Complete\")\n",
    "\n",
    "        return df_train, df_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ Data processing ] Scale action data, Add noise, Round data, clip data \n",
      "[ Data processing ] Split train and test data \n",
      "[ Data processing ] Complete\n"
     ]
    }
   ],
   "source": [
    "clsDataProcessor = Data_processor()\n",
    "objTrans_train = objTransitions(**params)\n",
    "objTrans_test = objTransitions(**params)\n",
    "objTrans_train.df, objTrans_test.df = clsDataProcessor.main(objTrans.df, **params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fe472eb3b80>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe472ea3ee0>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe472e3f130>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe472e2ee20>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cur_act, cur_obs = objTrans_train.get_cur_transitions()\n",
    "cur_act.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fe472cd7bb0>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe472cd37f0>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe472cf2310>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe472c98790>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cur_obs[acc_feature_list].hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fe472bb2640>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe472b8e220>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fe472b49c10>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fe472b7b0d0>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cur_obs[vel_feature_list].hist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Unit Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "test_clip_scaled_df (__main__.TestDataProcessor) ... ok\n",
      "test_duplicate_Noise_df (__main__.TestDataProcessor) ... ok\n",
      "test_duplicate_target_df (__main__.TestDataProcessor) ... FAIL\n",
      "test_scale_df_act (__main__.TestDataProcessor) ... ok\n",
      "test_scale_df_obs (__main__.TestDataProcessor) ... ok\n",
      "\n",
      "======================================================================\n",
      "FAIL: test_duplicate_target_df (__main__.TestDataProcessor)\n",
      "----------------------------------------------------------------------\n",
      "Traceback (most recent call last):\n",
      "  File \"<ipython-input-42-3a802cd8e44a>\", line 119, in test_duplicate_target_df\n",
      "    assert_frame_equal(results, example)\n",
      "  File \"/home/yliu2/anaconda3/envs/py3/lib/python3.8/site-packages/pandas/_testing.py\", line 1561, in assert_frame_equal\n",
      "    raise_assert_detail(\n",
      "  File \"/home/yliu2/anaconda3/envs/py3/lib/python3.8/site-packages/pandas/_testing.py\", line 1036, in raise_assert_detail\n",
      "    raise AssertionError(msg)\n",
      "AssertionError: DataFrame are different\n",
      "\n",
      "DataFrame shape mismatch\n",
      "[left]:  (6, 8)\n",
      "[right]: (6, 5)\n",
      "\n",
      "----------------------------------------------------------------------\n",
      "Ran 5 tests in 0.023s\n",
      "\n",
      "FAILED (failures=1)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<unittest.main.TestProgram at 0x7fe472a56130>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class TestDataProcessor(unittest.TestCase):\n",
    "    def __init__(self, *args, **kwargs):\n",
    "        super().__init__(*args, **kwargs)\n",
    "\n",
    "        self.clsDataProcessor = Data_processor()\n",
    "        self.act_scale_bnd = {\n",
    "            \"act_range\": (1000, 2000),\n",
    "            \"scaled_act_range\": (-1, 1),\n",
    "        }\n",
    "\n",
    "    def test_scale_df_act(self):\n",
    "        test_df = pd.DataFrame(\n",
    "            {\n",
    "                \"df\": [1, 2, 3],\n",
    "                \"m0\": [1000, 1500, 2000],\n",
    "                \"m1\": [100, 200, 300],\n",
    "                \"m2\": [3000, 5000, 10000],\n",
    "            }\n",
    "        )\n",
    "        act_filter_feature = [\"m0\", \"m1\", \"m2\"]\n",
    "\n",
    "        example = pd.DataFrame(\n",
    "            {\n",
    "                \"df\": [1, 2, 3],\n",
    "                \"m0\": [-1.0, 0.0, 1.0],\n",
    "                \"m1\": [-2.8, -2.6, -2.4],\n",
    "                \"m2\": [3.0, 7.0, 17.0],\n",
    "            }\n",
    "        )\n",
    "\n",
    "        results = self.clsDataProcessor.scale_df_act(\n",
    "            test_df, self.act_scale_bnd, act_filter_feature\n",
    "        )\n",
    "        assert_frame_equal(results, example)\n",
    "\n",
    "    def test_scale_df_obs(self):\n",
    "        test_df = pd.DataFrame(\n",
    "            {\n",
    "                \"Time\": [0.1, 0.2, 0.3],\n",
    "                \"linear_acceleration.x\": [-8, 18, 5],\n",
    "                \"orientation.w\": [0, 1, 0.5],\n",
    "                \"angVelocity.y\": [-20, 60, 40],\n",
    "                \"velocity.linear.z\": [-4, 8, 2],\n",
    "                \"target.velocity.z\": [-4, 8, 2],\n",
    "                \"m0\": [1500, 2000, 1000],\n",
    "            }\n",
    "        )\n",
    "\n",
    "        obs_range_dic = {\n",
    "            \"linear_acceleration.x\": (-8, 18),\n",
    "            \"orientation.w\": (0, 1),\n",
    "            \"angVelocity.y\": (-20, 60),\n",
    "            \"velocity.linear.z\": (-4, 8),\n",
    "            \"target.velocity.z\": (-4, 8),\n",
    "        }\n",
    "        obs_filter_feature = [\n",
    "            \"linear_acceleration.x\",\n",
    "            \"orientation.w\",\n",
    "            \"angVelocity.y\",\n",
    "            \"velocity.linear.z\",\n",
    "            \"target.velocity.z\",\n",
    "        ]\n",
    "        obs_scale_bnd = {\"scaled_obs_range\": (-1, 1)}\n",
    "        for key in obs_filter_feature:\n",
    "            obs_scale_bnd[key] = obs_range_dic[key]\n",
    "\n",
    "        example = pd.DataFrame(\n",
    "            {\n",
    "                \"Time\": [0.1, 0.2, 0.3],\n",
    "                \"linear_acceleration.x\": [-1.0, 1.0, 0],\n",
    "                \"orientation.w\": [-1.0, 1.0, 0],\n",
    "                \"angVelocity.y\": [-1.0, 1.0, 0.5],\n",
    "                \"velocity.linear.z\": [-1.0, 1.0, 0],\n",
    "                \"target.velocity.z\": [-1.0, 1.0, 0],\n",
    "                \"m0\": [1500, 2000, 1000],\n",
    "            }\n",
    "        )\n",
    "\n",
    "        results = self.clsDataProcessor.scale_df_obs(\n",
    "            test_df, obs_scale_bnd, obs_filter_feature\n",
    "        )\n",
    "        assert_frame_equal(results, example)\n",
    "\n",
    "    def test_duplicate_target_df(self):\n",
    "        test_df = pd.DataFrame(\n",
    "            {\n",
    "                \"Time\": [0.1, 0.2],\n",
    "                \"df\": [1500, 2000],\n",
    "                \"t0\": [3.5, -4.2],\n",
    "                \"t1\": [2.1, -3.9],\n",
    "                \"t2\": [1.0, -2.5],\n",
    "            }\n",
    "        )\n",
    "\n",
    "        num_target_df = 2\n",
    "        target_range_list = [(-7, 7), (-7, 7), (-1, 1)]\n",
    "        target_freq_range_list = [(0.05, 0.001), (0.05, 0.001), (0.1, 0.001)]\n",
    "        target_feature_list = [\"t0\", \"t1\", \"t2\"]\n",
    "        target_dict = {\n",
    "            \"num_target\": num_target_df,\n",
    "            \"target_feature_list\": target_feature_list,\n",
    "            \"range_list\": target_range_list,\n",
    "            \"freq_range_list\": target_freq_range_list,\n",
    "        }\n",
    "\n",
    "        example = pd.DataFrame(\n",
    "            {\n",
    "                \"Time\": [0.1, 0.2, 0.1, 0.2, 0.1, 0.2],\n",
    "                \"df\": [1500, 2000, 1500, 2000, 1500, 2000],\n",
    "                \"t0\": [3.5, -4.2, -2.835839, -2.825216, 0.458193, 0.459025],\n",
    "                \"t1\": [2.1, -3.9, 2.427205, 2.444603, -2.597170, -2.586763],\n",
    "                \"t2\": [1.0, -2.5, -0.865367, -0.866228, -0.115274, -0.119987],\n",
    "            }\n",
    "        )\n",
    "\n",
    "        results = self.clsDataProcessor.duplicate_target_df(\n",
    "            test_df, **target_dict, seed=123\n",
    "        )\n",
    "        assert_frame_equal(results, example)\n",
    "\n",
    "    def test_duplicate_Noise_df(self):\n",
    "        test_df = pd.DataFrame(\n",
    "            {\"Time\": [0, 1], \"f1\": [0, 0], \"f2\": [1000, 2000], \"f3\": [-1000, -999]}\n",
    "        )\n",
    "        num_noise_df = 2\n",
    "        noise_level = 0.5\n",
    "\n",
    "        example = pd.DataFrame(\n",
    "            {\n",
    "                \"Time\": [0, 1, 0, 1, 0, 1],\n",
    "                \"f1\": [0.0, 0.0, 0.0, 0.0, 0.0, 0.0],\n",
    "                \"f2\": [1000, 2000, 893.069667, 2109.734485, 1092.414869, 1921.589008],\n",
    "                \"f3\": [\n",
    "                    -1000,\n",
    "                    -999,\n",
    "                    -1000.136574,\n",
    "                    -999.038447,\n",
    "                    -1000.009534,\n",
    "                    -998.885475,\n",
    "                ],\n",
    "            }\n",
    "        )\n",
    "\n",
    "        results = self.clsDataProcessor.duplicate_Noise_df(\n",
    "            test_df, num_noise_df, noise_level, seed=123\n",
    "        )\n",
    "        assert_frame_equal(results, example)\n",
    "\n",
    "    def test_clip_scaled_df(self):\n",
    "        test_df = pd.DataFrame(\n",
    "            {\n",
    "                \"df\": [1, 2, 3],\n",
    "                \"m0\": [-1.0, 0.0, 1.0],\n",
    "                \"m1\": [0.0, -2.6, -2.4],\n",
    "                \"m2\": [1.0, 7.0, 17.0],\n",
    "            }\n",
    "        )\n",
    "        act_filter_feature = [\"m0\", \"m1\", \"m2\"]\n",
    "\n",
    "        example = pd.DataFrame({\"df\": [1], \"m0\": [-1.0], \"m1\": [0.0], \"m2\": [1.0]})\n",
    "\n",
    "        results = self.clsDataProcessor.clip_scaled_df(\n",
    "            test_df, self.act_scale_bnd, act_filter_feature\n",
    "        )\n",
    "        assert_frame_equal(results, example)\n",
    "\n",
    "\n",
    "unittest.main(argv=[\"\"], verbosity=2, exit=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Design Reward Function"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Reward Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "class RewardFn:\n",
    "    def reward_fn(\n",
    "        self,\n",
    "        next_obs_df,\n",
    "        prev_act_list,\n",
    "        prev_obs_list,\n",
    "        obs_reward_weights=(1, 1),\n",
    "        reward_weights=(1, 0, 1, 1),\n",
    "        **params\n",
    "    ):\n",
    "\n",
    "        next_obsCost_np = 0\n",
    "        next_obsCost_np, _ = self.calculate_obsCost(\n",
    "            next_obs_df, obs_reward_weights, **params\n",
    "        )\n",
    "\n",
    "        prev_obsCost_np = 0\n",
    "        for obs_df in prev_obs_list:\n",
    "            obsCost_np, obs_cost_info = self.calculate_obsCost(\n",
    "                obs_df, obs_reward_weights, **params\n",
    "            )\n",
    "            prev_obsCost_np += obsCost_np\n",
    "\n",
    "        prev_actCost_np = 0\n",
    "        for act_df in prev_act_list:\n",
    "            actCost_np = self.calculate_actCost(act_df, actCost_feature_list)\n",
    "            prev_actCost_np += actCost_np\n",
    "\n",
    "        prev_actChangeCost_np = 0\n",
    "        for i in range(len(prev_act_list) - 1):\n",
    "            act_df, next_act_df = prev_act_list[i], prev_act_list[i + 1]\n",
    "            actChangeCost_np = self.calculate_actChangeCost(\n",
    "                act_df, next_act_df, actChangeCost_feature_list\n",
    "            )\n",
    "            prev_actChangeCost_np += actChangeCost_np\n",
    "\n",
    "        rewards = (\n",
    "            reward_weights[0] * next_obsCost_np\n",
    "            + reward_weights[1] * prev_obsCost_np\n",
    "            + reward_weights[2] * prev_actCost_np\n",
    "            + reward_weights[3] * prev_actChangeCost_np\n",
    "        )\n",
    "\n",
    "        rewards = pd.DataFrame({\"rewards\": rewards})\n",
    "        info = pd.DataFrame(\n",
    "            {\n",
    "                \"next_obsCost\": next_obsCost_np,\n",
    "                \"prev_obsCost\": prev_obsCost_np,\n",
    "                \"prev_actCost\": prev_actCost_np,\n",
    "                \"prev_actChangeCost\": prev_actChangeCost_np,\n",
    "            }\n",
    "        )\n",
    "        return rewards, info, obs_cost_info\n",
    "\n",
    "    def calculate_obsCost(self, obs_df, obs_reward_weights, **params):\n",
    "        xy_target_df = obs_df[xy_target_feature_list]\n",
    "        xy_cost_df = obs_df[xy_cost_feature_list]\n",
    "        z_target_df = obs_df[z_target_feature_list]\n",
    "        z_cost_df = obs_df[z_cost_feature_list]\n",
    "        acc_cost_df = obs_df[acc_feature_list]\n",
    "\n",
    "        xyCost_np = self.calculate_xyCost(xy_cost_df, xy_target_df)\n",
    "        zCost_np = self.calculate_zCost(z_cost_df, z_target_df)\n",
    "        accCost_np = self.calculate_accCost(acc_cost_df)\n",
    "\n",
    "        obs_cost_np = (\n",
    "            obs_reward_weights[0] * xyCost_np\n",
    "            + obs_reward_weights[1] * zCost_np\n",
    "            + obs_reward_weights[2] * accCost_np\n",
    "        )\n",
    "        obs_cost_info = pd.DataFrame(\n",
    "            {\n",
    "                \"xyCost_np\": xyCost_np,\n",
    "                \"zCost_np\": zCost_np,\n",
    "                \"accCost_np\": accCost_np,\n",
    "            }\n",
    "        )\n",
    "        return obs_cost_np, obs_cost_info\n",
    "\n",
    "    def calculate_xyCost(self, xy_df, xy_target_df):\n",
    "        xyCost_np = xy_target_df.to_numpy() - xy_df.to_numpy()\n",
    "        xyCost_np = -np.linalg.norm(xyCost_np, axis=1)\n",
    "        return xyCost_np\n",
    "\n",
    "    def calculate_zCost(self, z_df, z_target_df):\n",
    "        zCost_np = z_target_df.to_numpy() - z_df.to_numpy()\n",
    "        zCost_np = -np.abs(zCost_np).reshape(\n",
    "            zCost_np.shape[0],\n",
    "        )\n",
    "        return zCost_np\n",
    "\n",
    "    def calculate_accCost(self, acc_cost_df):\n",
    "        accCost_np = acc_cost_df.to_numpy()\n",
    "        accCost_np[:, 2] = accCost_np[:, 2] - acceleration_constant\n",
    "        accCost_np = -np.linalg.norm(accCost_np, axis=1)\n",
    "        return accCost_np\n",
    "\n",
    "    def calculate_actCost(self, act_df, actCost_feature_list):\n",
    "        actCost_np = act_df[actCost_feature_list].to_numpy()\n",
    "        actCost_np = -np.linalg.norm(actCost_np, axis=1)\n",
    "        return actCost_np\n",
    "\n",
    "    def calculate_actChangeCost(self, act_df, next_act_df, actChangeCost_feature_list):\n",
    "        actChangeCost_np = (\n",
    "            next_act_df[actChangeCost_feature_list].to_numpy()\n",
    "            - act_df[actChangeCost_feature_list].to_numpy()\n",
    "        )\n",
    "        actChangeCost_np = -np.linalg.norm(actChangeCost_np, axis=1)\n",
    "        return actChangeCost_np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Post Processor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Post_processor:\n",
    "    def post_processing(self, objTrans, **params):\n",
    "        df = objTrans.df\n",
    "        new_params = params\n",
    "        if bool_merge_target_vel:\n",
    "            df = self.merge_diff_feature(\n",
    "                df, target_feature_list, vel_feature_list, diff_vel_name\n",
    "            )\n",
    "\n",
    "            new_params[\"obs_filter_feature\"] = diff_vel_filter_feature\n",
    "\n",
    "        new_objTrans = objTransitions(**new_params)\n",
    "        new_objTrans.df = df\n",
    "        return new_objTrans\n",
    "\n",
    "    def merge_diff_feature(self, df, feature_list_a, feature_list_b, new_name):\n",
    "        df_a = df[feature_list_a]\n",
    "        df_b = df[feature_list_b]\n",
    "\n",
    "        df_a.columns = new_name\n",
    "        df_b.columns = new_name\n",
    "        diff = df_a - df_b\n",
    "\n",
    "        df = df.drop(feature_list_a, axis=1)\n",
    "        df = df.drop(feature_list_b, axis=1)\n",
    "        df = pd.concat([df, diff], axis=1)\n",
    "        return df\n",
    "\n",
    "    def split_xy(self, prev_act_list, prev_obs_list, next_obs, rewards, **params):\n",
    "        x_df_list = []\n",
    "        for obs_df, act_df in zip(prev_obs_list, prev_act_list):\n",
    "            obs_act_df = pd.concat([obs_df, act_df], axis=1)\n",
    "            x_df_list.append(obs_act_df)\n",
    "        x = pd.concat(x_df_list, axis=1)\n",
    "        y = pd.concat([rewards, next_obs], axis=1)\n",
    "        return x, y\n",
    "\n",
    "    def main(self, objTrans, **params):\n",
    "        prev_act_list, prev_obs_list = objTrans.get_prev_transitions_list()\n",
    "        _next_act, next_obs = objTrans.get_next_transitions()\n",
    "\n",
    "        clsRewardFn = RewardFn()\n",
    "        rewards, info, obs_cost_info = clsRewardFn.reward_fn(\n",
    "            next_obs, prev_act_list, prev_obs_list, **params\n",
    "        )\n",
    "\n",
    "        objTrans = self.post_processing(objTrans, **params)\n",
    "        prev_act_list, prev_obs_list = objTrans.get_prev_transitions_list()\n",
    "        _next_act, next_obs = objTrans.get_next_transitions()\n",
    "\n",
    "        x, y = self.split_xy(prev_act_list, prev_obs_list, next_obs, rewards, **params)\n",
    "\n",
    "        return x, y, objTrans, rewards, info, obs_cost_info"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "clsPostProcessor = Post_processor()\n",
    "(\n",
    "    x_train,\n",
    "    y_train,\n",
    "    objTrans_post_Train,\n",
    "    rewards_train,\n",
    "    info_train,\n",
    "    obsCost_info_train,\n",
    ") = clsPostProcessor.main(objTrans_train, **params)\n",
    "(\n",
    "    x_test,\n",
    "    y_test,\n",
    "    objTrans_post_Test,\n",
    "    rewards_test,\n",
    "    info_test,\n",
    "    obsCost_info_test,\n",
    ") = clsPostProcessor.main(objTrans_test, **params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Time</th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>...</th>\n",
       "      <th>stick</th>\n",
       "      <th>m1</th>\n",
       "      <th>unused0</th>\n",
       "      <th>m0</th>\n",
       "      <th>unused1</th>\n",
       "      <th>unused2</th>\n",
       "      <th>unused3</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>40.720</td>\n",
       "      <td>-0.339639</td>\n",
       "      <td>0.110418</td>\n",
       "      <td>-0.827665</td>\n",
       "      <td>-0.000766</td>\n",
       "      <td>0.077680</td>\n",
       "      <td>0.001966</td>\n",
       "      <td>0.999453</td>\n",
       "      <td>-0.000940</td>\n",
       "      <td>-0.465561</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1500.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.005457</td>\n",
       "      <td>0.000061</td>\n",
       "      <td>-0.010283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>40.750</td>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1500.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>40.855</td>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1500.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>40.950</td>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1500.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>41.055</td>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1500.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130047</th>\n",
       "      <td>530.255</td>\n",
       "      <td>-0.365871</td>\n",
       "      <td>0.110524</td>\n",
       "      <td>-0.839970</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.061592</td>\n",
       "      <td>-0.279552</td>\n",
       "      <td>0.938296</td>\n",
       "      <td>-0.027287</td>\n",
       "      <td>-0.498203</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.743448</td>\n",
       "      <td>1609.977041</td>\n",
       "      <td>984.818384</td>\n",
       "      <td>0.243784</td>\n",
       "      <td>997.265879</td>\n",
       "      <td>1016.259325</td>\n",
       "      <td>987.747875</td>\n",
       "      <td>-0.111981</td>\n",
       "      <td>0.016982</td>\n",
       "      <td>-0.019615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130048</th>\n",
       "      <td>530.350</td>\n",
       "      <td>-0.341890</td>\n",
       "      <td>0.133136</td>\n",
       "      <td>-0.840814</td>\n",
       "      <td>0.013045</td>\n",
       "      <td>0.069504</td>\n",
       "      <td>-0.326998</td>\n",
       "      <td>0.929452</td>\n",
       "      <td>-0.016595</td>\n",
       "      <td>-0.487728</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.728756</td>\n",
       "      <td>1625.631849</td>\n",
       "      <td>992.327502</td>\n",
       "      <td>0.250594</td>\n",
       "      <td>990.558742</td>\n",
       "      <td>975.447146</td>\n",
       "      <td>1010.251659</td>\n",
       "      <td>-0.113057</td>\n",
       "      <td>0.015711</td>\n",
       "      <td>0.000852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130049</th>\n",
       "      <td>530.455</td>\n",
       "      <td>-0.343339</td>\n",
       "      <td>0.092339</td>\n",
       "      <td>-0.833314</td>\n",
       "      <td>0.053400</td>\n",
       "      <td>0.052476</td>\n",
       "      <td>-0.300955</td>\n",
       "      <td>0.961942</td>\n",
       "      <td>-0.029044</td>\n",
       "      <td>-0.494937</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.703249</td>\n",
       "      <td>1604.175507</td>\n",
       "      <td>1008.266036</td>\n",
       "      <td>0.207016</td>\n",
       "      <td>990.431086</td>\n",
       "      <td>998.697455</td>\n",
       "      <td>988.931005</td>\n",
       "      <td>-0.123274</td>\n",
       "      <td>-0.052128</td>\n",
       "      <td>0.002326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130050</th>\n",
       "      <td>530.550</td>\n",
       "      <td>-0.359142</td>\n",
       "      <td>0.113944</td>\n",
       "      <td>-0.832440</td>\n",
       "      <td>0.053497</td>\n",
       "      <td>0.076679</td>\n",
       "      <td>-0.232153</td>\n",
       "      <td>0.894869</td>\n",
       "      <td>-0.028039</td>\n",
       "      <td>-0.500465</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.707815</td>\n",
       "      <td>1615.722845</td>\n",
       "      <td>1002.958931</td>\n",
       "      <td>0.228232</td>\n",
       "      <td>978.719945</td>\n",
       "      <td>986.488988</td>\n",
       "      <td>978.123875</td>\n",
       "      <td>-0.119334</td>\n",
       "      <td>-0.008596</td>\n",
       "      <td>0.005163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130051</th>\n",
       "      <td>530.655</td>\n",
       "      <td>-0.360915</td>\n",
       "      <td>0.123887</td>\n",
       "      <td>-0.841583</td>\n",
       "      <td>0.045156</td>\n",
       "      <td>0.064502</td>\n",
       "      <td>-0.324552</td>\n",
       "      <td>0.923933</td>\n",
       "      <td>-0.032147</td>\n",
       "      <td>-0.503145</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.714354</td>\n",
       "      <td>1610.057602</td>\n",
       "      <td>999.390513</td>\n",
       "      <td>0.246285</td>\n",
       "      <td>1024.146703</td>\n",
       "      <td>1021.701050</td>\n",
       "      <td>975.383439</td>\n",
       "      <td>-0.177710</td>\n",
       "      <td>-0.066836</td>\n",
       "      <td>-0.021639</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>130052 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Time  linear_acceleration.x  linear_acceleration.y  \\\n",
       "0        40.720              -0.339639               0.110418   \n",
       "1        40.750              -0.338895               0.110588   \n",
       "2        40.855              -0.337344               0.111285   \n",
       "3        40.950              -0.336520               0.110705   \n",
       "4        41.055              -0.337113               0.110990   \n",
       "...         ...                    ...                    ...   \n",
       "130047  530.255              -0.365871               0.110524   \n",
       "130048  530.350              -0.341890               0.133136   \n",
       "130049  530.455              -0.343339               0.092339   \n",
       "130050  530.550              -0.359142               0.113944   \n",
       "130051  530.655              -0.360915               0.123887   \n",
       "\n",
       "        linear_acceleration.z  orientation.x  orientation.y  orientation.z  \\\n",
       "0                   -0.827665      -0.000766       0.077680       0.001966   \n",
       "1                   -0.827918      -0.000749       0.078977       0.001969   \n",
       "2                   -0.829133      -0.000721       0.080688       0.001974   \n",
       "3                   -0.829874      -0.000679       0.084465       0.001989   \n",
       "4                   -0.829482      -0.000653       0.085655       0.001995   \n",
       "...                       ...            ...            ...            ...   \n",
       "130047              -0.839970       0.011774       0.061592      -0.279552   \n",
       "130048              -0.840814       0.013045       0.069504      -0.326998   \n",
       "130049              -0.833314       0.053400       0.052476      -0.300955   \n",
       "130050              -0.832440       0.053497       0.076679      -0.232153   \n",
       "130051              -0.841583       0.045156       0.064502      -0.324552   \n",
       "\n",
       "        orientation.w  angVelocity.x  angVelocity.y  ...     stick  \\\n",
       "0            0.999453      -0.000940      -0.465561  ...  0.000000   \n",
       "1            0.999435      -0.000537      -0.468295  ...  0.000000   \n",
       "2            0.999410      -0.000495      -0.469974  ...  0.000000   \n",
       "3            0.999354      -0.000737      -0.477887  ...  0.000000   \n",
       "4            0.999336      -0.000583      -0.482817  ...  0.000000   \n",
       "...               ...            ...            ...  ...       ...   \n",
       "130047       0.938296      -0.027287      -0.498203  ... -0.743448   \n",
       "130048       0.929452      -0.016595      -0.487728  ... -0.728756   \n",
       "130049       0.961942      -0.029044      -0.494937  ... -0.703249   \n",
       "130050       0.894869      -0.028039      -0.500465  ... -0.707815   \n",
       "130051       0.923933      -0.032147      -0.503145  ... -0.714354   \n",
       "\n",
       "                 m1      unused0        m0      unused1      unused2  \\\n",
       "0       1500.000000     0.000000  0.000000     0.000000     0.000000   \n",
       "1       1500.000000     0.000000  0.000000     0.000000     0.000000   \n",
       "2       1500.000000     0.000000  0.000000     0.000000     0.000000   \n",
       "3       1500.000000     0.000000  0.000000     0.000000     0.000000   \n",
       "4       1500.000000     0.000000  0.000000     0.000000     0.000000   \n",
       "...             ...          ...       ...          ...          ...   \n",
       "130047  1609.977041   984.818384  0.243784   997.265879  1016.259325   \n",
       "130048  1625.631849   992.327502  0.250594   990.558742   975.447146   \n",
       "130049  1604.175507  1008.266036  0.207016   990.431086   998.697455   \n",
       "130050  1615.722845  1002.958931  0.228232   978.719945   986.488988   \n",
       "130051  1610.057602   999.390513  0.246285  1024.146703  1021.701050   \n",
       "\n",
       "            unused3  diff_vel.x  diff_vel.y  diff_vel.z  \n",
       "0          0.000000    0.005457    0.000061   -0.010283  \n",
       "1          0.000000    0.004691    0.000042   -0.009506  \n",
       "2          0.000000    0.003962    0.000032   -0.008575  \n",
       "3          0.000000    0.003315    0.000030   -0.007324  \n",
       "4          0.000000    0.002794    0.000020   -0.005945  \n",
       "...             ...         ...         ...         ...  \n",
       "130047   987.747875   -0.111981    0.016982   -0.019615  \n",
       "130048  1010.251659   -0.113057    0.015711    0.000852  \n",
       "130049   988.931005   -0.123274   -0.052128    0.002326  \n",
       "130050   978.123875   -0.119334   -0.008596    0.005163  \n",
       "130051   975.383439   -0.177710   -0.066836   -0.021639  \n",
       "\n",
       "[130052 rows x 26 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "objTrans_post_Train.df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>angVelocity.z</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.339639</td>\n",
       "      <td>0.110418</td>\n",
       "      <td>-0.827665</td>\n",
       "      <td>-0.000766</td>\n",
       "      <td>0.077680</td>\n",
       "      <td>0.001966</td>\n",
       "      <td>0.999453</td>\n",
       "      <td>-0.000940</td>\n",
       "      <td>-0.465561</td>\n",
       "      <td>0.001066</td>\n",
       "      <td>0.005457</td>\n",
       "      <td>0.000061</td>\n",
       "      <td>-0.010283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>0.001074</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>0.001201</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>0.001150</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130046</th>\n",
       "      <td>-0.344923</td>\n",
       "      <td>0.098582</td>\n",
       "      <td>-0.843070</td>\n",
       "      <td>0.053583</td>\n",
       "      <td>0.060754</td>\n",
       "      <td>-0.266596</td>\n",
       "      <td>0.872743</td>\n",
       "      <td>-0.020147</td>\n",
       "      <td>-0.487226</td>\n",
       "      <td>0.027103</td>\n",
       "      <td>-0.170622</td>\n",
       "      <td>0.003831</td>\n",
       "      <td>-0.019575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130047</th>\n",
       "      <td>-0.365871</td>\n",
       "      <td>0.110524</td>\n",
       "      <td>-0.839970</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.061592</td>\n",
       "      <td>-0.279552</td>\n",
       "      <td>0.938296</td>\n",
       "      <td>-0.027287</td>\n",
       "      <td>-0.498203</td>\n",
       "      <td>-0.036114</td>\n",
       "      <td>-0.111981</td>\n",
       "      <td>0.016982</td>\n",
       "      <td>-0.019615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130048</th>\n",
       "      <td>-0.341890</td>\n",
       "      <td>0.133136</td>\n",
       "      <td>-0.840814</td>\n",
       "      <td>0.013045</td>\n",
       "      <td>0.069504</td>\n",
       "      <td>-0.326998</td>\n",
       "      <td>0.929452</td>\n",
       "      <td>-0.016595</td>\n",
       "      <td>-0.487728</td>\n",
       "      <td>0.029309</td>\n",
       "      <td>-0.113057</td>\n",
       "      <td>0.015711</td>\n",
       "      <td>0.000852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130049</th>\n",
       "      <td>-0.343339</td>\n",
       "      <td>0.092339</td>\n",
       "      <td>-0.833314</td>\n",
       "      <td>0.053400</td>\n",
       "      <td>0.052476</td>\n",
       "      <td>-0.300955</td>\n",
       "      <td>0.961942</td>\n",
       "      <td>-0.029044</td>\n",
       "      <td>-0.494937</td>\n",
       "      <td>-0.019181</td>\n",
       "      <td>-0.123274</td>\n",
       "      <td>-0.052128</td>\n",
       "      <td>0.002326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130050</th>\n",
       "      <td>-0.359142</td>\n",
       "      <td>0.113944</td>\n",
       "      <td>-0.832440</td>\n",
       "      <td>0.053497</td>\n",
       "      <td>0.076679</td>\n",
       "      <td>-0.232153</td>\n",
       "      <td>0.894869</td>\n",
       "      <td>-0.028039</td>\n",
       "      <td>-0.500465</td>\n",
       "      <td>-0.029371</td>\n",
       "      <td>-0.119334</td>\n",
       "      <td>-0.008596</td>\n",
       "      <td>0.005163</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>130051 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        linear_acceleration.x  linear_acceleration.y  linear_acceleration.z  \\\n",
       "0                   -0.339639               0.110418              -0.827665   \n",
       "1                   -0.338895               0.110588              -0.827918   \n",
       "2                   -0.337344               0.111285              -0.829133   \n",
       "3                   -0.336520               0.110705              -0.829874   \n",
       "4                   -0.337113               0.110990              -0.829482   \n",
       "...                       ...                    ...                    ...   \n",
       "130046              -0.344923               0.098582              -0.843070   \n",
       "130047              -0.365871               0.110524              -0.839970   \n",
       "130048              -0.341890               0.133136              -0.840814   \n",
       "130049              -0.343339               0.092339              -0.833314   \n",
       "130050              -0.359142               0.113944              -0.832440   \n",
       "\n",
       "        orientation.x  orientation.y  orientation.z  orientation.w  \\\n",
       "0           -0.000766       0.077680       0.001966       0.999453   \n",
       "1           -0.000749       0.078977       0.001969       0.999435   \n",
       "2           -0.000721       0.080688       0.001974       0.999410   \n",
       "3           -0.000679       0.084465       0.001989       0.999354   \n",
       "4           -0.000653       0.085655       0.001995       0.999336   \n",
       "...               ...            ...            ...            ...   \n",
       "130046       0.053583       0.060754      -0.266596       0.872743   \n",
       "130047       0.011774       0.061592      -0.279552       0.938296   \n",
       "130048       0.013045       0.069504      -0.326998       0.929452   \n",
       "130049       0.053400       0.052476      -0.300955       0.961942   \n",
       "130050       0.053497       0.076679      -0.232153       0.894869   \n",
       "\n",
       "        angVelocity.x  angVelocity.y  angVelocity.z  diff_vel.x  diff_vel.y  \\\n",
       "0           -0.000940      -0.465561       0.001066    0.005457    0.000061   \n",
       "1           -0.000537      -0.468295       0.001074    0.004691    0.000042   \n",
       "2           -0.000495      -0.469974       0.001091    0.003962    0.000032   \n",
       "3           -0.000737      -0.477887       0.001201    0.003315    0.000030   \n",
       "4           -0.000583      -0.482817       0.001150    0.002794    0.000020   \n",
       "...               ...            ...            ...         ...         ...   \n",
       "130046      -0.020147      -0.487226       0.027103   -0.170622    0.003831   \n",
       "130047      -0.027287      -0.498203      -0.036114   -0.111981    0.016982   \n",
       "130048      -0.016595      -0.487728       0.029309   -0.113057    0.015711   \n",
       "130049      -0.029044      -0.494937      -0.019181   -0.123274   -0.052128   \n",
       "130050      -0.028039      -0.500465      -0.029371   -0.119334   -0.008596   \n",
       "\n",
       "        diff_vel.z  \n",
       "0        -0.010283  \n",
       "1        -0.009506  \n",
       "2        -0.008575  \n",
       "3        -0.007324  \n",
       "4        -0.005945  \n",
       "...            ...  \n",
       "130046   -0.019575  \n",
       "130047   -0.019615  \n",
       "130048    0.000852  \n",
       "130049    0.002326  \n",
       "130050    0.005163  \n",
       "\n",
       "[130051 rows x 13 columns]"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cur_act, cur_obs = objTrans_post_Train.get_cur_transitions()\n",
    "cur_obs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>angVelocity.z</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "      <th>lfin</th>\n",
       "      <th>tfin</th>\n",
       "      <th>stick</th>\n",
       "      <th>m0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.339639</td>\n",
       "      <td>0.110418</td>\n",
       "      <td>-0.827665</td>\n",
       "      <td>-0.000766</td>\n",
       "      <td>0.077680</td>\n",
       "      <td>0.001966</td>\n",
       "      <td>0.999453</td>\n",
       "      <td>-0.000940</td>\n",
       "      <td>-0.465561</td>\n",
       "      <td>0.001066</td>\n",
       "      <td>0.005457</td>\n",
       "      <td>0.000061</td>\n",
       "      <td>-0.010283</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>0.001074</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>0.001201</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>0.001150</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130046</th>\n",
       "      <td>-0.344923</td>\n",
       "      <td>0.098582</td>\n",
       "      <td>-0.843070</td>\n",
       "      <td>0.053583</td>\n",
       "      <td>0.060754</td>\n",
       "      <td>-0.266596</td>\n",
       "      <td>0.872743</td>\n",
       "      <td>-0.020147</td>\n",
       "      <td>-0.487226</td>\n",
       "      <td>0.027103</td>\n",
       "      <td>-0.170622</td>\n",
       "      <td>0.003831</td>\n",
       "      <td>-0.019575</td>\n",
       "      <td>0.085505</td>\n",
       "      <td>0.025398</td>\n",
       "      <td>-0.752155</td>\n",
       "      <td>0.209998</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130047</th>\n",
       "      <td>-0.365871</td>\n",
       "      <td>0.110524</td>\n",
       "      <td>-0.839970</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.061592</td>\n",
       "      <td>-0.279552</td>\n",
       "      <td>0.938296</td>\n",
       "      <td>-0.027287</td>\n",
       "      <td>-0.498203</td>\n",
       "      <td>-0.036114</td>\n",
       "      <td>-0.111981</td>\n",
       "      <td>0.016982</td>\n",
       "      <td>-0.019615</td>\n",
       "      <td>0.119570</td>\n",
       "      <td>0.026084</td>\n",
       "      <td>-0.743448</td>\n",
       "      <td>0.243784</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130048</th>\n",
       "      <td>-0.341890</td>\n",
       "      <td>0.133136</td>\n",
       "      <td>-0.840814</td>\n",
       "      <td>0.013045</td>\n",
       "      <td>0.069504</td>\n",
       "      <td>-0.326998</td>\n",
       "      <td>0.929452</td>\n",
       "      <td>-0.016595</td>\n",
       "      <td>-0.487728</td>\n",
       "      <td>0.029309</td>\n",
       "      <td>-0.113057</td>\n",
       "      <td>0.015711</td>\n",
       "      <td>0.000852</td>\n",
       "      <td>0.118006</td>\n",
       "      <td>0.038970</td>\n",
       "      <td>-0.728756</td>\n",
       "      <td>0.250594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130049</th>\n",
       "      <td>-0.343339</td>\n",
       "      <td>0.092339</td>\n",
       "      <td>-0.833314</td>\n",
       "      <td>0.053400</td>\n",
       "      <td>0.052476</td>\n",
       "      <td>-0.300955</td>\n",
       "      <td>0.961942</td>\n",
       "      <td>-0.029044</td>\n",
       "      <td>-0.494937</td>\n",
       "      <td>-0.019181</td>\n",
       "      <td>-0.123274</td>\n",
       "      <td>-0.052128</td>\n",
       "      <td>0.002326</td>\n",
       "      <td>0.081308</td>\n",
       "      <td>0.045735</td>\n",
       "      <td>-0.703249</td>\n",
       "      <td>0.207016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130050</th>\n",
       "      <td>-0.359142</td>\n",
       "      <td>0.113944</td>\n",
       "      <td>-0.832440</td>\n",
       "      <td>0.053497</td>\n",
       "      <td>0.076679</td>\n",
       "      <td>-0.232153</td>\n",
       "      <td>0.894869</td>\n",
       "      <td>-0.028039</td>\n",
       "      <td>-0.500465</td>\n",
       "      <td>-0.029371</td>\n",
       "      <td>-0.119334</td>\n",
       "      <td>-0.008596</td>\n",
       "      <td>0.005163</td>\n",
       "      <td>0.082342</td>\n",
       "      <td>0.037605</td>\n",
       "      <td>-0.707815</td>\n",
       "      <td>0.228232</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>130051 rows × 17 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        linear_acceleration.x  linear_acceleration.y  linear_acceleration.z  \\\n",
       "0                   -0.339639               0.110418              -0.827665   \n",
       "1                   -0.338895               0.110588              -0.827918   \n",
       "2                   -0.337344               0.111285              -0.829133   \n",
       "3                   -0.336520               0.110705              -0.829874   \n",
       "4                   -0.337113               0.110990              -0.829482   \n",
       "...                       ...                    ...                    ...   \n",
       "130046              -0.344923               0.098582              -0.843070   \n",
       "130047              -0.365871               0.110524              -0.839970   \n",
       "130048              -0.341890               0.133136              -0.840814   \n",
       "130049              -0.343339               0.092339              -0.833314   \n",
       "130050              -0.359142               0.113944              -0.832440   \n",
       "\n",
       "        orientation.x  orientation.y  orientation.z  orientation.w  \\\n",
       "0           -0.000766       0.077680       0.001966       0.999453   \n",
       "1           -0.000749       0.078977       0.001969       0.999435   \n",
       "2           -0.000721       0.080688       0.001974       0.999410   \n",
       "3           -0.000679       0.084465       0.001989       0.999354   \n",
       "4           -0.000653       0.085655       0.001995       0.999336   \n",
       "...               ...            ...            ...            ...   \n",
       "130046       0.053583       0.060754      -0.266596       0.872743   \n",
       "130047       0.011774       0.061592      -0.279552       0.938296   \n",
       "130048       0.013045       0.069504      -0.326998       0.929452   \n",
       "130049       0.053400       0.052476      -0.300955       0.961942   \n",
       "130050       0.053497       0.076679      -0.232153       0.894869   \n",
       "\n",
       "        angVelocity.x  angVelocity.y  angVelocity.z  diff_vel.x  diff_vel.y  \\\n",
       "0           -0.000940      -0.465561       0.001066    0.005457    0.000061   \n",
       "1           -0.000537      -0.468295       0.001074    0.004691    0.000042   \n",
       "2           -0.000495      -0.469974       0.001091    0.003962    0.000032   \n",
       "3           -0.000737      -0.477887       0.001201    0.003315    0.000030   \n",
       "4           -0.000583      -0.482817       0.001150    0.002794    0.000020   \n",
       "...               ...            ...            ...         ...         ...   \n",
       "130046      -0.020147      -0.487226       0.027103   -0.170622    0.003831   \n",
       "130047      -0.027287      -0.498203      -0.036114   -0.111981    0.016982   \n",
       "130048      -0.016595      -0.487728       0.029309   -0.113057    0.015711   \n",
       "130049      -0.029044      -0.494937      -0.019181   -0.123274   -0.052128   \n",
       "130050      -0.028039      -0.500465      -0.029371   -0.119334   -0.008596   \n",
       "\n",
       "        diff_vel.z      lfin      tfin     stick        m0  \n",
       "0        -0.010283  0.000000  0.000000  0.000000  0.000000  \n",
       "1        -0.009506  0.000000  0.000000  0.000000  0.000000  \n",
       "2        -0.008575  0.000000  0.000000  0.000000  0.000000  \n",
       "3        -0.007324  0.000000  0.000000  0.000000  0.000000  \n",
       "4        -0.005945  0.000000  0.000000  0.000000  0.000000  \n",
       "...            ...       ...       ...       ...       ...  \n",
       "130046   -0.019575  0.085505  0.025398 -0.752155  0.209998  \n",
       "130047   -0.019615  0.119570  0.026084 -0.743448  0.243784  \n",
       "130048    0.000852  0.118006  0.038970 -0.728756  0.250594  \n",
       "130049    0.002326  0.081308  0.045735 -0.703249  0.207016  \n",
       "130050    0.005163  0.082342  0.037605 -0.707815  0.228232  \n",
       "\n",
       "[130051 rows x 17 columns]"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rewards</th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>angVelocity.z</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.147286</td>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>0.001074</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.146053</td>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.145019</td>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>0.001201</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.144760</td>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>0.001150</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.144314</td>\n",
       "      <td>-0.338472</td>\n",
       "      <td>0.111655</td>\n",
       "      <td>-0.828684</td>\n",
       "      <td>-0.000575</td>\n",
       "      <td>0.087582</td>\n",
       "      <td>0.002002</td>\n",
       "      <td>0.999305</td>\n",
       "      <td>-0.000491</td>\n",
       "      <td>-0.488700</td>\n",
       "      <td>0.001007</td>\n",
       "      <td>0.001747</td>\n",
       "      <td>-0.000004</td>\n",
       "      <td>-0.002957</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130046</th>\n",
       "      <td>-0.348912</td>\n",
       "      <td>-0.365871</td>\n",
       "      <td>0.110524</td>\n",
       "      <td>-0.839970</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.061592</td>\n",
       "      <td>-0.279552</td>\n",
       "      <td>0.938296</td>\n",
       "      <td>-0.027287</td>\n",
       "      <td>-0.498203</td>\n",
       "      <td>-0.036114</td>\n",
       "      <td>-0.111981</td>\n",
       "      <td>0.016982</td>\n",
       "      <td>-0.019615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130047</th>\n",
       "      <td>-0.357394</td>\n",
       "      <td>-0.341890</td>\n",
       "      <td>0.133136</td>\n",
       "      <td>-0.840814</td>\n",
       "      <td>0.013045</td>\n",
       "      <td>0.069504</td>\n",
       "      <td>-0.326998</td>\n",
       "      <td>0.929452</td>\n",
       "      <td>-0.016595</td>\n",
       "      <td>-0.487728</td>\n",
       "      <td>0.029309</td>\n",
       "      <td>-0.113057</td>\n",
       "      <td>0.015711</td>\n",
       "      <td>0.000852</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130048</th>\n",
       "      <td>-0.365734</td>\n",
       "      <td>-0.343339</td>\n",
       "      <td>0.092339</td>\n",
       "      <td>-0.833314</td>\n",
       "      <td>0.053400</td>\n",
       "      <td>0.052476</td>\n",
       "      <td>-0.300955</td>\n",
       "      <td>0.961942</td>\n",
       "      <td>-0.029044</td>\n",
       "      <td>-0.494937</td>\n",
       "      <td>-0.019181</td>\n",
       "      <td>-0.123274</td>\n",
       "      <td>-0.052128</td>\n",
       "      <td>0.002326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130049</th>\n",
       "      <td>-0.352857</td>\n",
       "      <td>-0.359142</td>\n",
       "      <td>0.113944</td>\n",
       "      <td>-0.832440</td>\n",
       "      <td>0.053497</td>\n",
       "      <td>0.076679</td>\n",
       "      <td>-0.232153</td>\n",
       "      <td>0.894869</td>\n",
       "      <td>-0.028039</td>\n",
       "      <td>-0.500465</td>\n",
       "      <td>-0.029371</td>\n",
       "      <td>-0.119334</td>\n",
       "      <td>-0.008596</td>\n",
       "      <td>0.005163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130050</th>\n",
       "      <td>-0.421707</td>\n",
       "      <td>-0.360915</td>\n",
       "      <td>0.123887</td>\n",
       "      <td>-0.841583</td>\n",
       "      <td>0.045156</td>\n",
       "      <td>0.064502</td>\n",
       "      <td>-0.324552</td>\n",
       "      <td>0.923933</td>\n",
       "      <td>-0.032147</td>\n",
       "      <td>-0.503145</td>\n",
       "      <td>0.039107</td>\n",
       "      <td>-0.177710</td>\n",
       "      <td>-0.066836</td>\n",
       "      <td>-0.021639</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>130051 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         rewards  linear_acceleration.x  linear_acceleration.y  \\\n",
       "0      -0.147286              -0.338895               0.110588   \n",
       "1      -0.146053              -0.337344               0.111285   \n",
       "2      -0.145019              -0.336520               0.110705   \n",
       "3      -0.144760              -0.337113               0.110990   \n",
       "4      -0.144314              -0.338472               0.111655   \n",
       "...          ...                    ...                    ...   \n",
       "130046 -0.348912              -0.365871               0.110524   \n",
       "130047 -0.357394              -0.341890               0.133136   \n",
       "130048 -0.365734              -0.343339               0.092339   \n",
       "130049 -0.352857              -0.359142               0.113944   \n",
       "130050 -0.421707              -0.360915               0.123887   \n",
       "\n",
       "        linear_acceleration.z  orientation.x  orientation.y  orientation.z  \\\n",
       "0                   -0.827918      -0.000749       0.078977       0.001969   \n",
       "1                   -0.829133      -0.000721       0.080688       0.001974   \n",
       "2                   -0.829874      -0.000679       0.084465       0.001989   \n",
       "3                   -0.829482      -0.000653       0.085655       0.001995   \n",
       "4                   -0.828684      -0.000575       0.087582       0.002002   \n",
       "...                       ...            ...            ...            ...   \n",
       "130046              -0.839970       0.011774       0.061592      -0.279552   \n",
       "130047              -0.840814       0.013045       0.069504      -0.326998   \n",
       "130048              -0.833314       0.053400       0.052476      -0.300955   \n",
       "130049              -0.832440       0.053497       0.076679      -0.232153   \n",
       "130050              -0.841583       0.045156       0.064502      -0.324552   \n",
       "\n",
       "        orientation.w  angVelocity.x  angVelocity.y  angVelocity.z  \\\n",
       "0            0.999435      -0.000537      -0.468295       0.001074   \n",
       "1            0.999410      -0.000495      -0.469974       0.001091   \n",
       "2            0.999354      -0.000737      -0.477887       0.001201   \n",
       "3            0.999336      -0.000583      -0.482817       0.001150   \n",
       "4            0.999305      -0.000491      -0.488700       0.001007   \n",
       "...               ...            ...            ...            ...   \n",
       "130046       0.938296      -0.027287      -0.498203      -0.036114   \n",
       "130047       0.929452      -0.016595      -0.487728       0.029309   \n",
       "130048       0.961942      -0.029044      -0.494937      -0.019181   \n",
       "130049       0.894869      -0.028039      -0.500465      -0.029371   \n",
       "130050       0.923933      -0.032147      -0.503145       0.039107   \n",
       "\n",
       "        diff_vel.x  diff_vel.y  diff_vel.z  \n",
       "0         0.004691    0.000042   -0.009506  \n",
       "1         0.003962    0.000032   -0.008575  \n",
       "2         0.003315    0.000030   -0.007324  \n",
       "3         0.002794    0.000020   -0.005945  \n",
       "4         0.001747   -0.000004   -0.002957  \n",
       "...            ...         ...         ...  \n",
       "130046   -0.111981    0.016982   -0.019615  \n",
       "130047   -0.113057    0.015711    0.000852  \n",
       "130048   -0.123274   -0.052128    0.002326  \n",
       "130049   -0.119334   -0.008596    0.005163  \n",
       "130050   -0.177710   -0.066836   -0.021639  \n",
       "\n",
       "[130051 rows x 14 columns]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fda6888b820>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rewards_train.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fda68790280>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fda68751df0>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fda6876f550>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fda6870f9d0>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "info_train.hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fda68647520>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fda6863e970>],\n",
       "       [<matplotlib.axes._subplots.AxesSubplot object at 0x7fda645a6130>,\n",
       "        <matplotlib.axes._subplots.AxesSubplot object at 0x7fda645c6820>]],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "obsCost_info_train.hist()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save Data to CSV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Save to Simulation Results Path: ./blimpSim/results_Z\n"
     ]
    }
   ],
   "source": [
    "if data_sim_not_real:\n",
    "    print(f\" Save to Simulation Results Path: {SimResults_path}\")\n",
    "    cur_act, cur_obs = objTrans_post_Train.get_cur_transitions()\n",
    "    cur_obs.to_csv(os.path.join(SimResults_path, \"obs\"), index=False)\n",
    "    cur_act.to_csv(os.path.join(SimResults_path, \"act\"), index=False)\n",
    "    x_train.to_csv(os.path.join(SimResults_path, \"x_train\"), index=False)\n",
    "    y_train.to_csv(os.path.join(SimResults_path, \"y_train\"), index=False)\n",
    "    x_test.to_csv(os.path.join(SimResults_path, \"x_test\"), index=False)\n",
    "    y_test.to_csv(os.path.join(SimResults_path, \"y_test\"), index=False)\n",
    "\n",
    "    act_read_test = pd.read_csv(os.path.join(SimResults_path, \"act\"))\n",
    "    obs_read_test = pd.read_csv(os.path.join(SimResults_path, \"obs\"))\n",
    "    x_train_read_test = pd.read_csv(os.path.join(SimResults_path, \"x_train\"))\n",
    "    y_train_read_test = pd.read_csv(os.path.join(SimResults_path, \"y_train\"))\n",
    "\n",
    "else:\n",
    "    print(f\" Save to Real Results Path: {RealResults_path}\")\n",
    "    cur_act, cur_obs = objTrans_post_Train.get_cur_transitions()\n",
    "    cur_obs.to_csv(os.path.join(RealResults_path, \"obs\"), index=False)\n",
    "    cur_act.to_csv(os.path.join(RealResults_path, \"act\"), index=False)\n",
    "    x_train.to_csv(os.path.join(RealResults_path, \"x_train\"), index=False)\n",
    "    y_train.to_csv(os.path.join(RealResults_path, \"y_train\"), index=False)\n",
    "    x_test.to_csv(os.path.join(RealResults_path, \"x_test\"), index=False)\n",
    "    y_test.to_csv(os.path.join(RealResults_path, \"y_test\"), index=False)\n",
    "\n",
    "    act_read_test = pd.read_csv(os.path.join(RealResults_path, \"act\"))\n",
    "    obs_read_test = pd.read_csv(os.path.join(RealResults_path, \"obs\"))\n",
    "    x_train_read_test = pd.read_csv(os.path.join(RealResults_path, \"x_train\"))\n",
    "    y_train_read_test = pd.read_csv(os.path.join(RealResults_path, \"y_train\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>lfin</th>\n",
       "      <th>tfin</th>\n",
       "      <th>stick</th>\n",
       "      <th>m0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424309</th>\n",
       "      <td>-0.035194</td>\n",
       "      <td>0.057400</td>\n",
       "      <td>-0.974563</td>\n",
       "      <td>0.512312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424310</th>\n",
       "      <td>-0.029639</td>\n",
       "      <td>0.001403</td>\n",
       "      <td>-0.965634</td>\n",
       "      <td>0.462399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424311</th>\n",
       "      <td>-0.011662</td>\n",
       "      <td>-0.024456</td>\n",
       "      <td>-0.040450</td>\n",
       "      <td>0.004881</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424312</th>\n",
       "      <td>-0.022210</td>\n",
       "      <td>0.070695</td>\n",
       "      <td>-0.953321</td>\n",
       "      <td>0.473733</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424313</th>\n",
       "      <td>-0.043363</td>\n",
       "      <td>0.065340</td>\n",
       "      <td>-0.975515</td>\n",
       "      <td>0.453316</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>424314 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            lfin      tfin     stick        m0\n",
       "0       0.000000  0.000000  0.000000  0.000000\n",
       "1       0.000000  0.000000  0.000000  0.000000\n",
       "2       0.000000  0.000000  0.000000  0.000000\n",
       "3       0.000000  0.000000  0.000000  0.000000\n",
       "4       0.000000  0.000000  0.000000  0.000000\n",
       "...          ...       ...       ...       ...\n",
       "424309 -0.035194  0.057400 -0.974563  0.512312\n",
       "424310 -0.029639  0.001403 -0.965634  0.462399\n",
       "424311 -0.011662 -0.024456 -0.040450  0.004881\n",
       "424312 -0.022210  0.070695 -0.953321  0.473733\n",
       "424313 -0.043363  0.065340 -0.975515  0.453316\n",
       "\n",
       "[424314 rows x 4 columns]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "act_read_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>angVelocity.z</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.339639</td>\n",
       "      <td>0.110418</td>\n",
       "      <td>-0.827665</td>\n",
       "      <td>-0.000766</td>\n",
       "      <td>0.077680</td>\n",
       "      <td>0.001966</td>\n",
       "      <td>0.999453</td>\n",
       "      <td>-0.000940</td>\n",
       "      <td>-0.465561</td>\n",
       "      <td>0.001066</td>\n",
       "      <td>0.005457</td>\n",
       "      <td>0.000061</td>\n",
       "      <td>-0.010283</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>0.001074</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>0.001201</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>0.001150</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424309</th>\n",
       "      <td>-0.438380</td>\n",
       "      <td>0.073303</td>\n",
       "      <td>-0.852311</td>\n",
       "      <td>0.215374</td>\n",
       "      <td>-0.033317</td>\n",
       "      <td>0.908661</td>\n",
       "      <td>-0.275289</td>\n",
       "      <td>-0.063510</td>\n",
       "      <td>-0.417113</td>\n",
       "      <td>0.036478</td>\n",
       "      <td>-0.214286</td>\n",
       "      <td>0.171382</td>\n",
       "      <td>0.001765</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424310</th>\n",
       "      <td>-0.445324</td>\n",
       "      <td>0.070800</td>\n",
       "      <td>-0.856367</td>\n",
       "      <td>0.167090</td>\n",
       "      <td>-0.028372</td>\n",
       "      <td>0.935118</td>\n",
       "      <td>-0.318809</td>\n",
       "      <td>-0.012798</td>\n",
       "      <td>-0.452840</td>\n",
       "      <td>0.039133</td>\n",
       "      <td>-0.138943</td>\n",
       "      <td>0.084257</td>\n",
       "      <td>-0.008969</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424311</th>\n",
       "      <td>-0.377031</td>\n",
       "      <td>0.109649</td>\n",
       "      <td>-0.845591</td>\n",
       "      <td>0.100147</td>\n",
       "      <td>-0.069905</td>\n",
       "      <td>0.922519</td>\n",
       "      <td>-0.318626</td>\n",
       "      <td>-0.041398</td>\n",
       "      <td>-0.433528</td>\n",
       "      <td>0.004389</td>\n",
       "      <td>-0.123573</td>\n",
       "      <td>0.129343</td>\n",
       "      <td>0.004335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424312</th>\n",
       "      <td>-0.429479</td>\n",
       "      <td>0.065852</td>\n",
       "      <td>-0.888331</td>\n",
       "      <td>0.092686</td>\n",
       "      <td>-0.081990</td>\n",
       "      <td>0.946247</td>\n",
       "      <td>-0.313921</td>\n",
       "      <td>-0.041979</td>\n",
       "      <td>-0.400530</td>\n",
       "      <td>-0.020914</td>\n",
       "      <td>-0.153315</td>\n",
       "      <td>0.088856</td>\n",
       "      <td>0.005801</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424313</th>\n",
       "      <td>-0.406149</td>\n",
       "      <td>0.067067</td>\n",
       "      <td>-0.809074</td>\n",
       "      <td>0.100974</td>\n",
       "      <td>-0.038671</td>\n",
       "      <td>0.923386</td>\n",
       "      <td>-0.306891</td>\n",
       "      <td>-0.013662</td>\n",
       "      <td>-0.401388</td>\n",
       "      <td>-0.009004</td>\n",
       "      <td>-0.089006</td>\n",
       "      <td>0.088070</td>\n",
       "      <td>-0.017570</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>424314 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        linear_acceleration.x  linear_acceleration.y  linear_acceleration.z  \\\n",
       "0                   -0.339639               0.110418              -0.827665   \n",
       "1                   -0.338895               0.110588              -0.827918   \n",
       "2                   -0.337344               0.111285              -0.829133   \n",
       "3                   -0.336520               0.110705              -0.829874   \n",
       "4                   -0.337113               0.110990              -0.829482   \n",
       "...                       ...                    ...                    ...   \n",
       "424309              -0.438380               0.073303              -0.852311   \n",
       "424310              -0.445324               0.070800              -0.856367   \n",
       "424311              -0.377031               0.109649              -0.845591   \n",
       "424312              -0.429479               0.065852              -0.888331   \n",
       "424313              -0.406149               0.067067              -0.809074   \n",
       "\n",
       "        orientation.x  orientation.y  orientation.z  orientation.w  \\\n",
       "0           -0.000766       0.077680       0.001966       0.999453   \n",
       "1           -0.000749       0.078977       0.001969       0.999435   \n",
       "2           -0.000721       0.080688       0.001974       0.999410   \n",
       "3           -0.000679       0.084465       0.001989       0.999354   \n",
       "4           -0.000653       0.085655       0.001995       0.999336   \n",
       "...               ...            ...            ...            ...   \n",
       "424309       0.215374      -0.033317       0.908661      -0.275289   \n",
       "424310       0.167090      -0.028372       0.935118      -0.318809   \n",
       "424311       0.100147      -0.069905       0.922519      -0.318626   \n",
       "424312       0.092686      -0.081990       0.946247      -0.313921   \n",
       "424313       0.100974      -0.038671       0.923386      -0.306891   \n",
       "\n",
       "        angVelocity.x  angVelocity.y  angVelocity.z  diff_vel.x  diff_vel.y  \\\n",
       "0           -0.000940      -0.465561       0.001066    0.005457    0.000061   \n",
       "1           -0.000537      -0.468295       0.001074    0.004691    0.000042   \n",
       "2           -0.000495      -0.469974       0.001091    0.003962    0.000032   \n",
       "3           -0.000737      -0.477887       0.001201    0.003315    0.000030   \n",
       "4           -0.000583      -0.482817       0.001150    0.002794    0.000020   \n",
       "...               ...            ...            ...         ...         ...   \n",
       "424309      -0.063510      -0.417113       0.036478   -0.214286    0.171382   \n",
       "424310      -0.012798      -0.452840       0.039133   -0.138943    0.084257   \n",
       "424311      -0.041398      -0.433528       0.004389   -0.123573    0.129343   \n",
       "424312      -0.041979      -0.400530      -0.020914   -0.153315    0.088856   \n",
       "424313      -0.013662      -0.401388      -0.009004   -0.089006    0.088070   \n",
       "\n",
       "        diff_vel.z  \n",
       "0        -0.010283  \n",
       "1        -0.009506  \n",
       "2        -0.008575  \n",
       "3        -0.007324  \n",
       "4        -0.005945  \n",
       "...            ...  \n",
       "424309    0.001765  \n",
       "424310   -0.008969  \n",
       "424311    0.004335  \n",
       "424312    0.005801  \n",
       "424313   -0.017570  \n",
       "\n",
       "[424314 rows x 13 columns]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "obs_read_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>angVelocity.z</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "      <th>lfin</th>\n",
       "      <th>tfin</th>\n",
       "      <th>stick</th>\n",
       "      <th>m0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.339639</td>\n",
       "      <td>0.110418</td>\n",
       "      <td>-0.827665</td>\n",
       "      <td>-0.000766</td>\n",
       "      <td>0.077680</td>\n",
       "      <td>0.001966</td>\n",
       "      <td>0.999453</td>\n",
       "      <td>-0.000940</td>\n",
       "      <td>-0.465561</td>\n",
       "      <td>0.001066</td>\n",
       "      <td>0.005457</td>\n",
       "      <td>0.000061</td>\n",
       "      <td>-0.010283</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>0.001074</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>0.001201</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>0.001150</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424309</th>\n",
       "      <td>-0.438380</td>\n",
       "      <td>0.073303</td>\n",
       "      <td>-0.852311</td>\n",
       "      <td>0.215374</td>\n",
       "      <td>-0.033317</td>\n",
       "      <td>0.908661</td>\n",
       "      <td>-0.275289</td>\n",
       "      <td>-0.063510</td>\n",
       "      <td>-0.417113</td>\n",
       "      <td>0.036478</td>\n",
       "      <td>-0.214286</td>\n",
       "      <td>0.171382</td>\n",
       "      <td>0.001765</td>\n",
       "      <td>-0.035194</td>\n",
       "      <td>0.057400</td>\n",
       "      <td>-0.974563</td>\n",
       "      <td>0.512312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424310</th>\n",
       "      <td>-0.445324</td>\n",
       "      <td>0.070800</td>\n",
       "      <td>-0.856367</td>\n",
       "      <td>0.167090</td>\n",
       "      <td>-0.028372</td>\n",
       "      <td>0.935118</td>\n",
       "      <td>-0.318809</td>\n",
       "      <td>-0.012798</td>\n",
       "      <td>-0.452840</td>\n",
       "      <td>0.039133</td>\n",
       "      <td>-0.138943</td>\n",
       "      <td>0.084257</td>\n",
       "      <td>-0.008969</td>\n",
       "      <td>-0.029639</td>\n",
       "      <td>0.001403</td>\n",
       "      <td>-0.965634</td>\n",
       "      <td>0.462399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424311</th>\n",
       "      <td>-0.377031</td>\n",
       "      <td>0.109649</td>\n",
       "      <td>-0.845591</td>\n",
       "      <td>0.100147</td>\n",
       "      <td>-0.069905</td>\n",
       "      <td>0.922519</td>\n",
       "      <td>-0.318626</td>\n",
       "      <td>-0.041398</td>\n",
       "      <td>-0.433528</td>\n",
       "      <td>0.004389</td>\n",
       "      <td>-0.123573</td>\n",
       "      <td>0.129343</td>\n",
       "      <td>0.004335</td>\n",
       "      <td>-0.011662</td>\n",
       "      <td>-0.024456</td>\n",
       "      <td>-0.040450</td>\n",
       "      <td>0.004881</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424312</th>\n",
       "      <td>-0.429479</td>\n",
       "      <td>0.065852</td>\n",
       "      <td>-0.888331</td>\n",
       "      <td>0.092686</td>\n",
       "      <td>-0.081990</td>\n",
       "      <td>0.946247</td>\n",
       "      <td>-0.313921</td>\n",
       "      <td>-0.041979</td>\n",
       "      <td>-0.400530</td>\n",
       "      <td>-0.020914</td>\n",
       "      <td>-0.153315</td>\n",
       "      <td>0.088856</td>\n",
       "      <td>0.005801</td>\n",
       "      <td>-0.022210</td>\n",
       "      <td>0.070695</td>\n",
       "      <td>-0.953321</td>\n",
       "      <td>0.473733</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424313</th>\n",
       "      <td>-0.406149</td>\n",
       "      <td>0.067067</td>\n",
       "      <td>-0.809074</td>\n",
       "      <td>0.100974</td>\n",
       "      <td>-0.038671</td>\n",
       "      <td>0.923386</td>\n",
       "      <td>-0.306891</td>\n",
       "      <td>-0.013662</td>\n",
       "      <td>-0.401388</td>\n",
       "      <td>-0.009004</td>\n",
       "      <td>-0.089006</td>\n",
       "      <td>0.088070</td>\n",
       "      <td>-0.017570</td>\n",
       "      <td>-0.043363</td>\n",
       "      <td>0.065340</td>\n",
       "      <td>-0.975515</td>\n",
       "      <td>0.453316</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>424314 rows × 17 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        linear_acceleration.x  linear_acceleration.y  linear_acceleration.z  \\\n",
       "0                   -0.339639               0.110418              -0.827665   \n",
       "1                   -0.338895               0.110588              -0.827918   \n",
       "2                   -0.337344               0.111285              -0.829133   \n",
       "3                   -0.336520               0.110705              -0.829874   \n",
       "4                   -0.337113               0.110990              -0.829482   \n",
       "...                       ...                    ...                    ...   \n",
       "424309              -0.438380               0.073303              -0.852311   \n",
       "424310              -0.445324               0.070800              -0.856367   \n",
       "424311              -0.377031               0.109649              -0.845591   \n",
       "424312              -0.429479               0.065852              -0.888331   \n",
       "424313              -0.406149               0.067067              -0.809074   \n",
       "\n",
       "        orientation.x  orientation.y  orientation.z  orientation.w  \\\n",
       "0           -0.000766       0.077680       0.001966       0.999453   \n",
       "1           -0.000749       0.078977       0.001969       0.999435   \n",
       "2           -0.000721       0.080688       0.001974       0.999410   \n",
       "3           -0.000679       0.084465       0.001989       0.999354   \n",
       "4           -0.000653       0.085655       0.001995       0.999336   \n",
       "...               ...            ...            ...            ...   \n",
       "424309       0.215374      -0.033317       0.908661      -0.275289   \n",
       "424310       0.167090      -0.028372       0.935118      -0.318809   \n",
       "424311       0.100147      -0.069905       0.922519      -0.318626   \n",
       "424312       0.092686      -0.081990       0.946247      -0.313921   \n",
       "424313       0.100974      -0.038671       0.923386      -0.306891   \n",
       "\n",
       "        angVelocity.x  angVelocity.y  angVelocity.z  diff_vel.x  diff_vel.y  \\\n",
       "0           -0.000940      -0.465561       0.001066    0.005457    0.000061   \n",
       "1           -0.000537      -0.468295       0.001074    0.004691    0.000042   \n",
       "2           -0.000495      -0.469974       0.001091    0.003962    0.000032   \n",
       "3           -0.000737      -0.477887       0.001201    0.003315    0.000030   \n",
       "4           -0.000583      -0.482817       0.001150    0.002794    0.000020   \n",
       "...               ...            ...            ...         ...         ...   \n",
       "424309      -0.063510      -0.417113       0.036478   -0.214286    0.171382   \n",
       "424310      -0.012798      -0.452840       0.039133   -0.138943    0.084257   \n",
       "424311      -0.041398      -0.433528       0.004389   -0.123573    0.129343   \n",
       "424312      -0.041979      -0.400530      -0.020914   -0.153315    0.088856   \n",
       "424313      -0.013662      -0.401388      -0.009004   -0.089006    0.088070   \n",
       "\n",
       "        diff_vel.z      lfin      tfin     stick        m0  \n",
       "0        -0.010283  0.000000  0.000000  0.000000  0.000000  \n",
       "1        -0.009506  0.000000  0.000000  0.000000  0.000000  \n",
       "2        -0.008575  0.000000  0.000000  0.000000  0.000000  \n",
       "3        -0.007324  0.000000  0.000000  0.000000  0.000000  \n",
       "4        -0.005945  0.000000  0.000000  0.000000  0.000000  \n",
       "...            ...       ...       ...       ...       ...  \n",
       "424309    0.001765 -0.035194  0.057400 -0.974563  0.512312  \n",
       "424310   -0.008969 -0.029639  0.001403 -0.965634  0.462399  \n",
       "424311    0.004335 -0.011662 -0.024456 -0.040450  0.004881  \n",
       "424312    0.005801 -0.022210  0.070695 -0.953321  0.473733  \n",
       "424313   -0.017570 -0.043363  0.065340 -0.975515  0.453316  \n",
       "\n",
       "[424314 rows x 17 columns]"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train_read_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rewards</th>\n",
       "      <th>linear_acceleration.x</th>\n",
       "      <th>linear_acceleration.y</th>\n",
       "      <th>linear_acceleration.z</th>\n",
       "      <th>orientation.x</th>\n",
       "      <th>orientation.y</th>\n",
       "      <th>orientation.z</th>\n",
       "      <th>orientation.w</th>\n",
       "      <th>angVelocity.x</th>\n",
       "      <th>angVelocity.y</th>\n",
       "      <th>angVelocity.z</th>\n",
       "      <th>diff_vel.x</th>\n",
       "      <th>diff_vel.y</th>\n",
       "      <th>diff_vel.z</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.099646</td>\n",
       "      <td>-0.338895</td>\n",
       "      <td>0.110588</td>\n",
       "      <td>-0.827918</td>\n",
       "      <td>-0.000749</td>\n",
       "      <td>0.078977</td>\n",
       "      <td>0.001969</td>\n",
       "      <td>0.999435</td>\n",
       "      <td>-0.000537</td>\n",
       "      <td>-0.468295</td>\n",
       "      <td>0.001074</td>\n",
       "      <td>0.004691</td>\n",
       "      <td>0.000042</td>\n",
       "      <td>-0.009506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.098777</td>\n",
       "      <td>-0.337344</td>\n",
       "      <td>0.111285</td>\n",
       "      <td>-0.829133</td>\n",
       "      <td>-0.000721</td>\n",
       "      <td>0.080688</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>0.999410</td>\n",
       "      <td>-0.000495</td>\n",
       "      <td>-0.469974</td>\n",
       "      <td>0.001091</td>\n",
       "      <td>0.003962</td>\n",
       "      <td>0.000032</td>\n",
       "      <td>-0.008575</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.097557</td>\n",
       "      <td>-0.336520</td>\n",
       "      <td>0.110705</td>\n",
       "      <td>-0.829874</td>\n",
       "      <td>-0.000679</td>\n",
       "      <td>0.084465</td>\n",
       "      <td>0.001989</td>\n",
       "      <td>0.999354</td>\n",
       "      <td>-0.000737</td>\n",
       "      <td>-0.477887</td>\n",
       "      <td>0.001201</td>\n",
       "      <td>0.003315</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>-0.007324</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.096167</td>\n",
       "      <td>-0.337113</td>\n",
       "      <td>0.110990</td>\n",
       "      <td>-0.829482</td>\n",
       "      <td>-0.000653</td>\n",
       "      <td>0.085655</td>\n",
       "      <td>0.001995</td>\n",
       "      <td>0.999336</td>\n",
       "      <td>-0.000583</td>\n",
       "      <td>-0.482817</td>\n",
       "      <td>0.001150</td>\n",
       "      <td>0.002794</td>\n",
       "      <td>0.000020</td>\n",
       "      <td>-0.005945</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.091858</td>\n",
       "      <td>-0.340192</td>\n",
       "      <td>0.110863</td>\n",
       "      <td>-0.828063</td>\n",
       "      <td>-0.000527</td>\n",
       "      <td>0.088324</td>\n",
       "      <td>0.002008</td>\n",
       "      <td>0.999294</td>\n",
       "      <td>-0.000646</td>\n",
       "      <td>-0.494788</td>\n",
       "      <td>0.001051</td>\n",
       "      <td>0.001111</td>\n",
       "      <td>-0.000008</td>\n",
       "      <td>-0.001652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424309</th>\n",
       "      <td>-0.361907</td>\n",
       "      <td>-0.445324</td>\n",
       "      <td>0.070800</td>\n",
       "      <td>-0.856367</td>\n",
       "      <td>0.167090</td>\n",
       "      <td>-0.028372</td>\n",
       "      <td>0.935118</td>\n",
       "      <td>-0.318809</td>\n",
       "      <td>-0.012798</td>\n",
       "      <td>-0.452840</td>\n",
       "      <td>0.039133</td>\n",
       "      <td>-0.138943</td>\n",
       "      <td>0.084257</td>\n",
       "      <td>-0.008969</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424310</th>\n",
       "      <td>-0.328766</td>\n",
       "      <td>-0.377031</td>\n",
       "      <td>0.109649</td>\n",
       "      <td>-0.845591</td>\n",
       "      <td>0.100147</td>\n",
       "      <td>-0.069905</td>\n",
       "      <td>0.922519</td>\n",
       "      <td>-0.318626</td>\n",
       "      <td>-0.041398</td>\n",
       "      <td>-0.433528</td>\n",
       "      <td>0.004389</td>\n",
       "      <td>-0.123573</td>\n",
       "      <td>0.129343</td>\n",
       "      <td>0.004335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424311</th>\n",
       "      <td>-0.107131</td>\n",
       "      <td>-0.429479</td>\n",
       "      <td>0.065852</td>\n",
       "      <td>-0.888331</td>\n",
       "      <td>0.092686</td>\n",
       "      <td>-0.081990</td>\n",
       "      <td>0.946247</td>\n",
       "      <td>-0.313921</td>\n",
       "      <td>-0.041979</td>\n",
       "      <td>-0.400530</td>\n",
       "      <td>-0.020914</td>\n",
       "      <td>-0.153315</td>\n",
       "      <td>0.088856</td>\n",
       "      <td>0.005801</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424312</th>\n",
       "      <td>-0.345214</td>\n",
       "      <td>-0.406149</td>\n",
       "      <td>0.067067</td>\n",
       "      <td>-0.809074</td>\n",
       "      <td>0.100974</td>\n",
       "      <td>-0.038671</td>\n",
       "      <td>0.923386</td>\n",
       "      <td>-0.306891</td>\n",
       "      <td>-0.013662</td>\n",
       "      <td>-0.401388</td>\n",
       "      <td>-0.009004</td>\n",
       "      <td>-0.089006</td>\n",
       "      <td>0.088070</td>\n",
       "      <td>-0.017570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424313</th>\n",
       "      <td>-0.351179</td>\n",
       "      <td>-0.425711</td>\n",
       "      <td>0.069189</td>\n",
       "      <td>-0.880704</td>\n",
       "      <td>0.111222</td>\n",
       "      <td>-0.027492</td>\n",
       "      <td>0.923473</td>\n",
       "      <td>-0.296983</td>\n",
       "      <td>-0.036818</td>\n",
       "      <td>-0.452546</td>\n",
       "      <td>0.058279</td>\n",
       "      <td>-0.131685</td>\n",
       "      <td>0.172587</td>\n",
       "      <td>0.026456</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>424314 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         rewards  linear_acceleration.x  linear_acceleration.y  \\\n",
       "0      -0.099646              -0.338895               0.110588   \n",
       "1      -0.098777              -0.337344               0.111285   \n",
       "2      -0.097557              -0.336520               0.110705   \n",
       "3      -0.096167              -0.337113               0.110990   \n",
       "4      -0.091858              -0.340192               0.110863   \n",
       "...          ...                    ...                    ...   \n",
       "424309 -0.361907              -0.445324               0.070800   \n",
       "424310 -0.328766              -0.377031               0.109649   \n",
       "424311 -0.107131              -0.429479               0.065852   \n",
       "424312 -0.345214              -0.406149               0.067067   \n",
       "424313 -0.351179              -0.425711               0.069189   \n",
       "\n",
       "        linear_acceleration.z  orientation.x  orientation.y  orientation.z  \\\n",
       "0                   -0.827918      -0.000749       0.078977       0.001969   \n",
       "1                   -0.829133      -0.000721       0.080688       0.001974   \n",
       "2                   -0.829874      -0.000679       0.084465       0.001989   \n",
       "3                   -0.829482      -0.000653       0.085655       0.001995   \n",
       "4                   -0.828063      -0.000527       0.088324       0.002008   \n",
       "...                       ...            ...            ...            ...   \n",
       "424309              -0.856367       0.167090      -0.028372       0.935118   \n",
       "424310              -0.845591       0.100147      -0.069905       0.922519   \n",
       "424311              -0.888331       0.092686      -0.081990       0.946247   \n",
       "424312              -0.809074       0.100974      -0.038671       0.923386   \n",
       "424313              -0.880704       0.111222      -0.027492       0.923473   \n",
       "\n",
       "        orientation.w  angVelocity.x  angVelocity.y  angVelocity.z  \\\n",
       "0            0.999435      -0.000537      -0.468295       0.001074   \n",
       "1            0.999410      -0.000495      -0.469974       0.001091   \n",
       "2            0.999354      -0.000737      -0.477887       0.001201   \n",
       "3            0.999336      -0.000583      -0.482817       0.001150   \n",
       "4            0.999294      -0.000646      -0.494788       0.001051   \n",
       "...               ...            ...            ...            ...   \n",
       "424309      -0.318809      -0.012798      -0.452840       0.039133   \n",
       "424310      -0.318626      -0.041398      -0.433528       0.004389   \n",
       "424311      -0.313921      -0.041979      -0.400530      -0.020914   \n",
       "424312      -0.306891      -0.013662      -0.401388      -0.009004   \n",
       "424313      -0.296983      -0.036818      -0.452546       0.058279   \n",
       "\n",
       "        diff_vel.x  diff_vel.y  diff_vel.z  \n",
       "0         0.004691    0.000042   -0.009506  \n",
       "1         0.003962    0.000032   -0.008575  \n",
       "2         0.003315    0.000030   -0.007324  \n",
       "3         0.002794    0.000020   -0.005945  \n",
       "4         0.001111   -0.000008   -0.001652  \n",
       "...            ...         ...         ...  \n",
       "424309   -0.138943    0.084257   -0.008969  \n",
       "424310   -0.123573    0.129343    0.004335  \n",
       "424311   -0.153315    0.088856    0.005801  \n",
       "424312   -0.089006    0.088070   -0.017570  \n",
       "424313   -0.131685    0.172587    0.026456  \n",
       "\n",
       "[424314 rows x 14 columns]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_read_test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Mix Real and Simulation data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "if mix_sim_real_data:\n",
    "    act_real = pd.read_csv(os.path.join(RealResults_path, \"act\"))\n",
    "    obs_real = pd.read_csv(os.path.join(RealResults_path, \"obs\"))\n",
    "    x_train_real = pd.read_csv(os.path.join(RealResults_path, \"x_train\"))\n",
    "    y_train_real = pd.read_csv(os.path.join(RealResults_path, \"y_train\"))\n",
    "    x_test_real = pd.read_csv(os.path.join(RealResults_path, \"x_test\"))\n",
    "    y_test_real = pd.read_csv(os.path.join(RealResults_path, \"y_test\"))\n",
    "\n",
    "    act_sim = pd.read_csv(os.path.join(SimResults_path, \"act\"))\n",
    "    obs_sim = pd.read_csv(os.path.join(SimResults_path, \"obs\"))\n",
    "    x_train_sim = pd.read_csv(os.path.join(SimResults_path, \"x_train\"))\n",
    "    y_train_sim = pd.read_csv(os.path.join(SimResults_path, \"y_train\"))\n",
    "    x_test_sim = pd.read_csv(os.path.join(SimResults_path, \"x_test\"))\n",
    "    y_test_sim = pd.read_csv(os.path.join(SimResults_path, \"y_test\"))\n",
    "\n",
    "    act_mix = pd.concat([act_sim, act_real], axis=0)\n",
    "    obs_mix = pd.concat([obs_sim, obs_real], axis=0)\n",
    "    x_train_mix = pd.concat([x_train_sim, x_train_real], axis=0)\n",
    "    y_train_mix = pd.concat([y_train_sim, y_train_real], axis=0)\n",
    "    x_test_mix = pd.concat([x_test_sim, x_test_real], axis=0)\n",
    "    y_test_mix = pd.concat([y_test_sim, y_test_real], axis=0)\n",
    "\n",
    "    obs_mix.to_csv(os.path.join(MixResults_path, \"obs\"), index=False)\n",
    "    act_mix.to_csv(os.path.join(MixResults_path, \"act\"), index=False)\n",
    "    x_train_mix.to_csv(os.path.join(MixResults_path, \"x_train\"), index=False)\n",
    "    y_train_mix.to_csv(os.path.join(MixResults_path, \"y_train\"), index=False)\n",
    "    x_test_mix.to_csv(os.path.join(MixResults_path, \"x_test\"), index=False)\n",
    "    y_test_mix.to_csv(os.path.join(MixResults_path, \"y_test\"), index=False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  },
  "toc-autonumbering": false,
  "toc-showcode": false,
  "toc-showmarkdowntxt": false
 },
 "nbformat": 4,
 "nbformat_minor": 4
}