Active Inference MCP is a project that demonstrates the implementation of active inference in the Mountain Car Problem (MCP) using Unity ML-Agents. The goal of this project is to showcase how active inference can be applied to solve reinforcement learning problems in a Unity environment.
The Mountain Car Problem is a classic reinforcement learning task in which an underpowered car must drive up a hill. The car's engine is not strong enough to directly climb the hill, so the car must learn to use its momentum by moving back and forth between the hills to reach the goal at the top. Active Inference MCP demonstrates how Active Inference can be applied to solve this problem.
To install and set up the project, follow the instructions in the Unity ML-Agents Installation Guide. This will help you set up the required dependencies and environment for running the project.
- Implementation of the Mountain Car Problem using Active Inference
- Visualization of the agent's learning progress
- Customizable environment and agent parameters for experimentation