AUTOMATING RETRO GAMES USING THE REINFROCEMENT LEARNING

TEAM NAME - TEAM Trex

• TEAM LEADER - AADHAAR KOUL (5th semester cse dept)
• TEAM MEMBER - ARJUN CHARAK (5th semester cse dept)
• TEAM MEMBER - SHOBIT KITCHLOO (5th semester cse dept)
• TEAM MEMBER - SIDDHARTH BHAWANI (5th semester cse dept)
• TEAM MEMBER - ANIL KUMAR (5th semester cse dept)

CLASS COORDINATOR - ARJUN PURI

Complete Automation of some of the most renouned game titles using the reinforcement learning.

• EQUAL CONTRIBUTION
• NO PLAGIARISM
• COPYRIGHT RESERVED TO MIET JAMMU 5TH SEMESTER CLASS A1 STUDENTS

ABSTRACT

General game testing relies on the use of human play testers, play test scripting, and prior knowledge of areas of interest to produce relevant test data. Using deep reinforcement learning (DRL), we introduce a self-learning mechanism to the game testing framework. With DRL, the framework is capable of exploring and/or exploiting the game mechanics based on a user-defined, reinforcing reward signal. As a result, test coverage is increased and unintended game play mechanics, exploits and bugs are discovered in a multitude of game types. In this paper, we show that DRL can be used to increase test coverage, find exploits, test map difficulty, and to detect common problems that arise in the testing of first-person shooter (FPS) games.In this paper, we study applying Reinforcement Learning to design a automatic agent to play the game Super Mario Bros. One of the challenge is how to handle the complex game environment. By abstracting the game environment into a state vector and using Q learning — an algorithm oblivious to transitional probabilities — we achieve tractable computation time and fast convergence. After training for 5000 iterations, our agent is able to win about 90 percent of the time. We also compare and analyze the choice of different learning rate (alpha) and discount factor (gamma)

Tech

• PYTHON - V3.10 , 3.8 AND 3.7
• PYTHON IDE

Specifications (Minimum requirement)

• RAM - 2GB
• PROCESSOR - CORE 2 DUO OR HIGHER FOR Q LEARNING TABLE PLOTTING
• WINDOWS VERSION - 10 OR HIGHER
• PYTHON INTERPRETER VERSION - PYTHON 3.7 AND PYTHON 10.0
• INTERNET CONNECTIVITY - NOT REQUIRED

NOTE : To install some of the packages the internet connectivity might be a requirement.

IMPLEMENTATION

The below given implementation is done for the snake game , so use the given below instructions for your benifit.

Open up the IDE command terminal . In our case we have used the VSCODE IDE and install the following packages:

• pygame

pip install pygame

• numpy

pip install numpy

• torch

pip install torch

• matplotlib

pip install matplotlib

• After the above are executed head on to the agent.py file and run the python code. A pygame window should appear with the snake game on it
• let it run and generate some of the punishment and rewards for the until it maintains a Q- table .
• At a certain point the snake should be able to get the rewards all the time with less errors
• At this stage the snake Ai will be trained to play the all on its own.

DOCUMENT AND THE PPT FILE CAN BE FOUND IN THIS REPOSITORY ONLY JUST CHECK THE REPOSITORY CONTENTS ABOVE

SAMPLES / PICTURES

First Contributions

This project aims to simplify and guide the way beginners make their first contribution. If you are looking to make your first contribution, follow the steps below.

If you're not comfortable with command line, here are tutorials using GUI tools.

If you don't have git on your machine, install it.

Fork this repository

Fork this repository by clicking on the fork button on the top of this page. This will create a copy of this repository in your account.

Clone the repository

Now clone the forked repository to your machine. Go to your GitHub account, open the forked repository, click on the code button and then click the copy to clipboard icon.

Open a terminal and run the following git command:

git clone "url you just copied"

where "url you just copied" (without the quotation marks) is the url to this repository (your fork of this project). See the previous steps to obtain the url.

For example:

git clone https://github.com/this-is-you/first-contributions.git

where this-is-you is your GitHub username. Here you're copying the contents of the first-contributions repository on GitHub to your computer.

Make necessary changes and commit those changes

Now open Contributors.md file in a text editor, add your name to it. Don't add it at the beginning or end of the file. Put it anywhere in between. Now, save the file.

If you go to the project directory and execute the command git status, you'll see there are changes.

Add those changes to the branch you just created using the git add command:

git add Contributors.md

Now commit those changes using the git commit command:

git commit -m "Add <your-name> to Contributors list"

replacing <your-name> with your name.

Push changes to GitHub

Push your changes using the command git push:

git push origin <add-your-branch-name>

replacing <add-your-branch-name> with the name of the branch you created earlier.

Submit your changes for review

If you go to your repository on GitHub, you'll see a Compare & pull request button. Click on that button.

Now submit the pull request.

Soon I'll be merging all your changes into the master branch of this project. You will get a notification email once the changes have been merged.

Where to go from here?

Congrats! You just completed the standard fork -> clone -> edit -> pull request workflow that you'll encounter often as a contributor!

Celebrate your contribution and share it with your friends and followers by going to web app.

You could join our slack team in case you need any help or have any questions. Join slack team.

Now let's get you started with contributing to other projects. We've compiled a list of projects with easy issues you can get started on. Check out the list of projects in the web app.

Additional material

Tutorials Using Other Tools

GitHub Desktop	Visual Studio 2017	GitKraken	Visual Studio Code	Atlassian Sourcetree	IntelliJ IDEA
添加内容

License

MIT

Free Software, Hell Yeah!