Minecraft Clone in Python and OpenGL to run Reinforcement Learning trainings. This project is W.I.P, many functionnalities are missing and the code is under optimized.
Control: WASD + LShift/Space + Mouse + Escape E - Save World F8 - Teleport to center of world Escape - Exit C - change camera (works only with agents)
- Base class for World Gen
- Two types of World Gen
- Dynamic World size
- Dynamic Seed for World Gen
- Customizable World Gen from functions (just like in the menu)
- Save & World and its entities
- Spawn agents in the center of the position
- Camera rotation needs to loop to avoid overflowing
- Test to lag the game to see if it doesn't break it
- Fix wall climbing
- Special agent blocks (spawn, marker, death, etc.)
- Rotation consistency between camera and agent
- Fix agent camera clipping through blocks
- Fix agent camera frustum
- Single Agent Handling
- Multi Agent Handling
- Agent 32x32 Camera with custom OpenGL settings
- RL processor for agents
- RL training steps
- RL mutations
- Base class for RL fitness
- Test types of RL fitness
- Saved world RL training and viewing settings
- RL training saving
- UI Interface
- Particles
