Optimization Algorithm for Feedback and Feedforward Policies towards Robot Control Robust to Sensing Failures

TL;DR

This paper introduces a unified optimization approach for feedback and feedforward policies in robot control, enhancing robustness to sensing failures by integrating control as inference with variational methods.

Contribution

It develops a novel optimization framework for jointly learning feedback and feedforward policies, improving robustness and stability in robotic control under sensing failures.

Findings

The method effectively optimizes combined policies in simulations and real robot experiments.

Feedforward policy demonstrates robustness to sensing failures, maintaining optimal motion.

The approach outperforms traditional reinforcement learning in stability and robustness.

Abstract

Model-free or learning-based control, in particular, reinforcement learning (RL), is expected to be applied for complex robotic tasks. Traditional RL requires a policy to be optimized is state-dependent, that means, the policy is a kind of feedback (FB) controllers. Due to the necessity of correct state observation in such a FB controller, it is sensitive to sensing failures. To alleviate this drawback of the FB controllers, feedback error learning integrates one of them with a feedforward (FF) controller. RL can be improved by dealing with the FB/FF policies, but to the best of our knowledge, a methodology for learning them in a unified manner has not been developed. In this paper, we propose a new optimization problem for optimizing both the FB/FF policies simultaneously. Inspired by control as inference, the optimization problem considers minimization/maximization of divergences between trajectory, predicted by the composed policy and a stochastic dynamics model, and optimal/non-optimal trajectories. By approximating the stochastic dynamics model using variational method, we naturally derive a regularization between the FB/FF policies. In numerical simulations and a robot experiment, we verified that the proposed method can stably optimize the composed policy even with the different learning law from the traditional RL. In addition, we demonstrated that the FF policy is robust to the sensing failures and can hold the optimal motion. Attached video is also uploaded on youtube: https://youtu.be/zLL4uXIRmrE