RoboLeague

A car soccer environment inspired by Rocket League for deep reinforcement learning experiments in an adversarial self-play setting.

Table of contest

• Installation
• Completed features (identical to Rocket League)
• Videos
• Planned features
• Useful reading
• Licence

Installation

Please refer to the Unity ML Agents repository. Current version of RoboLeague uses ML-Agents Release 11

Completed features (identical to Rocket League)

• Ball physics (ground / wall bounces, friction model, gravity, etc)
• Car aerial control and physics
• Acceleration and braking on the ground
• Single jump mechanics
• Sticky forces

Videos

Planned features

• OpenAI gym interface integration. Unity ML already provides a wrapper
• Car ball interaction
• Dodges and second jump
• Stabilizing moment near the ground
• Lateral friction model as in the game (ratio between Vx/Vy)

Useful reading (in random order)

• Unity ML-Agents Toolkit Documentation
• Physics and math of Rocket League
• Useful game values
• RLBot framework

Licence

Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)