This project was created for the Durham Academy ML/AI class
This project uses Pymunk physics simulation and Ray reinforcement learning to teach a ragdoll to run and play soccer. It uses a PPO model to control the ragdoll's joints and complete tasks, inspired by AI Warehouse on Youtube.
Each file is documented. Models are saved in the ./models folder. Training files are named train_(TASK).py (train_running.py and train_soccer.py). Training environments are saved as (TASK)_env.py and all other files contain PyGame environments for ragdoll/soccer field constructions. Visualize trained models with testing scrips saved as test_(TASK).py for both running and soccer.