active-inference-robotics

Synthesizes Patrick Kenny's discrete active inference framework with K-Scale's JAX/MuJoCo robotics stack for predictive coding in robot locomotion.

Use When

• Bridging active inference theory with robot control implementations
• Connecting KL divergence minimization to RL training loops
• Applying mean field approximation to robotics state estimation
• Implementing sim2real transfer as inference about future observations
• Training locomotion policies with principled entropy regularization

The Constructive Collision