Unbiased Active Inference for Classical Control

Active inference is a mathematical framework that originated in computational neuroscience. Recently, it has been demonstrated as a promising approach for constructing goal-driven behavior in robotics. Specifically, the active inference controller (AIC) has been successful on several continuous control and state-estimation tasks. Despite its relative success, some established design choices lead to a number of practical limitations for robot control. These include having a biased estimate of the state, and only an implicit model of control actions. In this paper, we highlight these limitations and propose an extended version of the unbiased active inference controller (u-AIC). The u-AIC maintains all the compelling benefits of the AIC and removes its limitations. Simulation results on a 2-DOF arm and experiments on a real 7-DOF manipulator show the improved performance of the u-AIC with respect to the standard AIC. The code can be found at https://github.com/cpezzato/unbiasedaic.

Authors

• Mohamed Baioumy
• Corrado Pezzato
• Riccardo M. G. Ferrari
• Nick Hawes

Keywords

• Simulation
• Robot control
• Computational neuroscience
• Human-robot interaction
• Manipulators
• State estimation
• Task analysis
• Active Inference
• Unbiased Inference
• Active Control
• Free Energy
• Gradient Descent
• Prediction Error
• Optimal Control
• Kalman Filter
• Sensor Locations
• Goal State
• Model Predictive Control
• Proportional-integral-derivative
• General Architecture
• Chain Rule
• Identity Mapping
• Rows Of Each Plot
• Feedforward Signal
• Fault-tolerant Control
• Sensor Precision
• Proof Of Convergence
• Real Robot
• Velocity Sensors
• Robotic Arm
• Scalar Case
• Reference Tracking
• Feedforward Control