beginning of action group graphing but not finished

learner_selector.py

@@ -23,7 +23,7 @@
 from rocket_learn.utils.stat_trackers.common_trackers import Speed, Demos, TimeoutRate, Touch, EpisodeLength, Boost, \
     BehindBall, TouchHeight, DistToBall, AirTouch, AirTouchHeight, BallHeight, BallSpeed, CarOnGround, GoalSpeed, \
     MaxGoalSpeed
-from my_stattrackers import GoalSpeedTop5perc, FlipReset
+from my_stattrackers import GoalSpeedTop5perc, FlipReset, ActionGroupingTracker
 from rlgym.utils.reward_functions.common_rewards import VelocityReward, EventReward
 from rlgym.utils.reward_functions.combined_reward import CombinedReward
 
@@ -55,10 +55,10 @@
         ent_coef=0.03,
     )
 
-    run_id = "selector_run_21.01"
+    run_id = "selector_run_test21.01"
     wandb.login(key=os.environ["WANDB_KEY"])
     logger = wandb.init(dir="./wandb_store",
-                        name="Selector_Run_21.01",
+                        name="Selector_Run_test21.01",
                         project="Opti",
                         entity="kaiyotech",
                         id=run_id,
@@ -147,6 +147,10 @@
                                         logger=logger,
                                         clear=False,
                                         stat_trackers=stat_trackers,
+                                        # action_grouping_tracker=ActionGroupingTracker(aerial_indices=[3, 6, 7, 8, 28, 29],
+                                        #                                               wall_indices=[8, 25, 26, 28, 29],
+                                        #                                               ground_indices=[0, 1, 2, 4, 5, *range(9, 25), 27, 29],
+                                        #                                               defend_indices=[3, 6, 7, 8, 28]),
                                         # gamemodes=("1v1", "2v2", "3v3"),
                                         max_age=1,
                                         pretrained_agents=Constants_selector.pretrained_agents

my_stattrackers.py

@@ -97,3 +97,42 @@ def update(self, gamestates: np.ndarray, mask: np.ndarray):
 
     def get_stat(self):
         return self.flip_reset_count / (self.count or 1)
+
+
+class ActionGroupingTracker:
+    # check if actions were used at the appropriate time
+
+    def __init__(self, aerial_indices, ground_indices, defend_indices, wall_indices):
+        self.name = "Action Grouping Tracker"
+        self.count = 0
+        self.flip_reset_count = 0
+
+    def reset(self):
+        self.count = 0
+        self.flip_reset_count = 0
+
+    def update(self, gamestates: np.ndarray, mask: np.ndarray):
+        players = gamestates[:, StateConstants.PLAYERS]
+        num_players = len(players[0]) // 39
+        has_jumps = players[:, StateConstants.HAS_JUMP]
+        # on_grounds = players[:, StateConstants.ON_GROUND]
+        players_x = players[:, StateConstants.CAR_POS_X]
+        players_y = players[:, StateConstants.CAR_POS_Y]
+        players_z = players[:, StateConstants.CAR_POS_Z]
+        for i in range(num_players):
+            has_jumps_player = has_jumps[:, i]
+            changes = np.where(has_jumps_player[:1] < has_jumps_player[:-1], True, False)
+            player_x = players_x[:, i]
+            player_y = players_y[:, i]
+            player_z = players_z[:, i]
+            on_grounds_player = np.where((player_z < 300) | (player_z > CEILING_Z - 300) |
+                                         ((-SIDE_WALL_X + 700) > player_x) |
+                                         ((SIDE_WALL_X - 700) > player_x) |
+                                         ((-BACK_WALL_Y + 700) > player_y) |
+                                         ((BACK_WALL_Y - 700) > player_y), True, False)
+            self.flip_reset_count += (~on_grounds_player[1:] & changes).sum()
+
+        self.count += has_jumps.size
+
+    def get_stat(self):
+        return self.flip_reset_count / (self.count or 1)