Scribd Logo
Upload

Welcome to Scribd!

  • 5 views

    CS 461 Project Proposal

    Uploaded by

    handeer22

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    CS 461 Project Proposal

    Uploaded by

    handeer22
    5 views2 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    5 views2 pages

    CS 461 Project Proposal

    Uploaded by

    handeer22

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    CS 461 Project Proposal

    Snake Game
    Emre Can Şen- 21902516, Ata Özlük - 21702335, Uygar Onat Erol - 21901908,
    Hande Eryılmaz - 21902678, Mete Ertan - 21903215

    Game Description
    Our agent will control a snake in this straightforward but compelling game, navigating it
    through a 40x40 grid to collect food items and subsequently extend its length. At the
    beginning of a game, the snake starts with a length of 3 and moves initially to the right. The
    food items are randomly positioned on the game board. The game's goal is to accumulate
    the greatest score while avoiding having the snake run into any walls or its own body.

    Gameplay
    ● Agent: In recent versions, the player usually moves the snake using arrow keys or
    touch controllers. The snake's body follows the head wherever it goes. In our game,
    the player will be the agent that we will be coding and the agent will try to accumulate
    the highest score possible.
    ● Collecting Food: The main objective of the game is to move around the playing
    board and eat the food that has been placed at various random spots. The snake
    lengthens as it consumes food. We will be adding poisonous food and it will make the
    snake smaller in size.
    ● Avoiding Obstacles: In order to accumulate a higher score in the game, the agent
    must avoid all obstacles, including the outer boundaries of the game board and its
    own body. The game ends when the agent collides with walls or its own body.
    ● Growing the Snake: The snake gets longer every time it consumes food. As the
    snake gets longer, the game gets harder since it's harder to move across the board
    without colliding with anything.

    Rules:
    ● Snake dies when it hits itself or the outer boundaries of the game board.
    ● If a snake eats poisonous food, its size becomes shorter.
    ● Only one non-poisonous food appears in the game screen at any time.
    ● Poisonous food will spawn in the game over time.
    ● The initial score is zero at the start of the game.

    Significance and Interesting Aspects


    The proposed project offers several significant and interesting aspects:
    ● Reinforcement learning in gaming: This research shows how reinforcement learning
    methods can be used in a traditional gaming setting that is both instructive and
    entertaining.
    ● Enhanced gameplay: The project introduces a new dimension to the original Snake
    game via the addition of poisonous food.
    Project Milestones
    To ensure steady progress throughout the project, we have outlined distinct goals to achieve:

    Milestone 1: Basic Snake Game Environment (Until Progress Report)


    ● Create a playable Snake game environment with a basic user interface.
    ● Implement snake movement, food generation, and collision detection.

    Milestone 2: User Interface and Documentation (Until Progress Report)


    ● Build a user interface to visualize the game progress and model's decisions.
    ● Begin documentation by creating code comments and an initial project report.

    Milestone 3: Reinforcement Learning Model


    ● Develop and train a reinforcement learning model for Snake using Q-learning and
    Deep Q-learning.
    ● Ensure the model can make decisions based on the current game state.

    Milestone 4: Enhanced Features


    ● Integrate enhanced features such as poisonous food.
    ● Adapt the reinforcement learning model to handle these new challenges.

    Milestone 5: Training and Evaluation


    ● Create a framework for training and evaluating the model's performance.
    ● Develop a reward system to encourage the model's learning process.

    Final Milestone: Project Completion


    ● Finalize the documentation, including a comprehensive project report.
    ● Conduct thorough testing and optimization to ensure the project's success.
    ● Prepare for a final project presentation and review.

    Conclusion
    The Reinforcement Learning Snake Game project attempts to include reinforcement learning
    while bringing novel elements to improve gameplay in a traditional gaming environment. We
    think this project has both amusing and informative elements, and it can be a useful learning
    experience for both users. The project will be successfully completed if the milestones
    outlined in this proposal are met. We are eager to get out on this adventure and investigate
    the fascinating interface between gaming and reinforcement learning.

    Previous Work
    ● Z. Wei, D. Wang, M. Zhang, A.-H. Tan, C. Miao, and Y. Zhou, “Autonomous Agents in Snake
    Game via Deep Reinforcement Learning,” IEEE Xplore, Jul. 01, 2018.
    https://ieeexplore.ieee.org/abstract/document/8460004

    You might also like