A Python recreation of the Walker project - a simulated environment for training populations of based robots using reinforcement learning evolutionary algorithms.
- Physics Simulation: Realistic 2D physics using Box2d
- Reinforcement Learning: Various implementations of Q-Learning
- Evolutionary Algorithm: Population-based evolution with genetic operators
- Real-time Visualization: Pygame-based rendering and debugging
- Interactive GUI: Parameter adjustment and simulation control
- Clone the repository:
git clone https://github.com/alecKarfonta/Walker
cd Walker- Build
docker compose up -d --buildRun the physics simulation test:
python run.pywalker_python/
├── src/
│ ├── config/ # Configuration and settings
│ ├── physics/ # Physics engine (Pymunk)
│ ├── agents/ # Reinforcement learning agents
│ ├── population/ # Population management
│ ├── rendering/ # Graphics and visualization
│ ├── ui/ # User interface
│ ├── utils/ # Utility functions
│ └── assets/ # Game assets
├── tests/ # Unit tests
├── requirements.txt # Python dependencies
└── run.py # Main entry point