CDS524 Learning Game Design

CDS524 Assignment 1 - Reinforcement Learning Game Design

Project Overview:
In this project, you will use Q-learning, a reinforcement learning algorithm, to design and implement your own game. The game can be any type of game you choose, from simple grid-based games to more complex simulations. The goal is to apply Q-learning to create a game where an agent (controlled by the algorithm) learns to make optimal decisions based on rewards and penalties.

Project Requirements:

1. Game Design:

• You must design a game with a clear objective and set of rules.
• The game should have a state space (all possible configurations of the game) and an action space (all possible actions the agent can take).
• The game should have a reward function that provides positive rewards for achieving the objective and negative rewards for undesirable actions or states.
  2. Q-Learning Implementation:
• You must implement the Q-learning algorithm in a 代写CDS524 Learning Game Designprogramming language you are comfortable with (e.g., Python).
• The algorithm should be able to learn a policy (a mapping from states to actions) that maximizes the cumulative reward over time.
• You should use techniques such as epsilon-greedy exploration, learning rate, and discount factor to balance exploration and exploitation.
  3. Game Interaction:
• You should create a user interface (UI) for your game that allows the agent to interact with the environment.
• The UI should display the current state of the game, the agent's actions, and any rewards or penalties received.
• You can use libraries or frameworks such as Pygame, Tkinterto create the UI and game environment.
  4. Documentation and Presentation:
• You should document your game design, Q-learning implementation, and evaluation results in a written report.
• You should also prepare a video presentation to demonstrate your game and explain the Q-learning algorithm, the challenges you faced, and the solutions you implemented.

Project Suggestions:

• Grid-Based Games: Design simple grid-based games such as Tic-Tac-Toe, the Grid World problem, or a maze-solving game.
• Board Games: Implement classic board games such as Chess, Checkers, or Go, and use Q-learning to create an AI opponent.
• Simulation Games: You can design simulation games such as a traffic simulation, a resource management game, or a city-building game.
• Action Games: For high grade wanted students, you can design action games such as a simple platformer or a shooting game, and use Q-learning to create an AI-controlled enemy or player character.

 

Rubric	Description	Marks
1. Game Design (10 marks)	Written	10
1.1	Clear objective and set of rules defined	2
1.2	State space and action space well-defined	2
1.3	Reward function correctly implemented with positive and negative rewards	3
1.4	Documentation of game design (including working diagrams if applicable)	3
2. Q-Learning Implementation (10 marks)	Written & Development	10
2.1	Q-learning algorithm implemented correctly in chosen programming language	3
2.2	Policy learned maximizes cumulative reward over time	3
2.3	Effective use of epsilon-greedy exploration, learning rate, and discount factor	2
2.4	Documentation and explanation of Q-learning implementation	2
3. Game Interaction (10 marks)	Development part	10
3.1	User interface (UI) allows agent to interact with environment	3
3.2	UI displays current state, agent's actions, and rewards/penalties	3
3.3	Appropriate use of libraries/frameworks for UI and game environment	2
3.4	User experience and interaction smooth and intuitive	2
4. Documentation and Presentation (10 marks)	Written & Presentation	10
4.1	Written report documenting game design, Q-learning implementation, and evaluation results	4
4.2	Video presentation demonstrating game and explaining Q-learning algorithm	3
4.3	Challenges faced and solutions implemented clearly explained in both report and presentation	2
4.4	Overall quality and clarity of documentation and presentation	1

 

Deliverables

1. A Google Colab (or equivalent) with your code and in-line explanations.
2. A brief report(1000-1500 words) evaluating the performance of your work.
3. Github repo
4. Youtube Demo Video

 

Submission Details: 

1. You must ensure that all your project files used for this task and the report sit in a directory called “Assignment 1 – Your Name”.
2. All files are required to be uploaded and a link to the “Assignment 1” directory submitted to Moodle.
3. Please make sure that unit Instructor and TA have access to the folder.
4. A linkto the demo video of your app running must be submitted.
5. It would be great if you could submit your GitHub
6. This is an individualassignment, and you should submit it by 8 pm, Tuesday, Week 6.