Adaptive Manipulator Control using Active Inference with Precision Learning

Active inference provides a framework for decisionmaking where the optimization is achieved by minimizing freeenergy. Previous work has used this framework for control and state-estimation of a robotic manipulator. This required manual definition of precision matrices which serve as controller gains. This paper provides an implementation for control and state-estimation where the precision matrices are tuned during execution-time (precision learning). Learning the precision matrices means automatically adjusting the controller’s gains which decreases oscillations and overshoot. Supplementary Material Code and further material is avaiable at: https://github.com/ MoBaioumy/active inference panda paper. Video: https://youtu.be/Ii1Ig1Lt0Xk