All Tools
U
AgentsFreeOpen Source
UNITY ML-AGENTS
Train RL and imitation learning agents inside Unity simulations
Apache-2.0
ABOUT
Training agents in photorealistic 3D worlds usually means writing a custom simulator and a separate RL stack. Unity ML-Agents connects Unity scenes to Python trainers so game NPCs, robots, and embodied agents can learn with PPO, SAC, or imitation learning without leaving the Unity editor.
INSTALL
pip install mlagentsINTEGRATION GUIDE
1. Train a game NPC or camera agent with PPO inside a Unity scene
2. Collect demonstrations in Unity and train an imitation-learning policy
3. Prototype robotics or warehouse sims in Unity then train with mlagents-learn
4. Export a trained behavior for runtime inference in a shipped Unity game
TAGS
pythonunityreinforcement-learningimitation-learningsimulationgames