AI4U is a multi-engine plugin (Godot and Unity) that allows you to design Non-Player Characters (NPCs) of games using an agent abstraction. In addition, AI4U has a low-level API that allows you to connect the agent to any algorithm made available in Python by the reinforcement learning community specifically and by the Artificial Intelligence community in general. Reinforcement learning promises to overcome traditional navigation mesh mechanisms in games and to provide more autonomous characters. AI4U can be integrated into Imitation Learning through Behavioral Cloning or Generative Adversarial Imitation Learning present on stable-baslines. Train using multiple concurrent Unity/Godot environment instances. Unity/Godot environment partial control from Python. Wrap Unity/Godot learning environments as a gym.
Features
- Support for multiple environment configurations and training scenarios
- Flexible SDK that can be integrated into your game or custom Unity scene
- API agnostic, but one provides support to all algorithms implemented by stable-basiles
- Integrated A3C implementation
- Tools for facilated no-markovian decision making
- Train robust agents using environment randomization