Robotic Control Optimization Through Kernel Selection in Safe Bayesian Optimization

TL;DR

This paper introduces a kernel selection approach within Safe Bayesian Optimization to improve control system optimization in high-dimensional robotics applications, demonstrated through drone PID tuning.

Contribution

It presents a novel kernel selection method that enhances Safe Bayesian Optimization for high-dimensional control problems, addressing scalability issues.

Findings

More efficient control optimization in high-dimensional systems

Significant improvements over existing methods in drone PID tuning

Validated on Safe Control Gym benchmark

Abstract

Control system optimization has long been a fundamental challenge in robotics. While recent advancements have led to the development of control algorithms that leverage learning-based approaches, such as SafeOpt, to optimize single feedback controllers, scaling these methods to high-dimensional complex systems with multiple controllers remains an open problem. In this paper, we propose a novel learning-based control optimization method, which enhances the additive Gaussian process-based Safe Bayesian Optimization algorithm to efficiently tackle high-dimensional problems through kernel selection. We use PID controller optimization in drones as a representative example and test the method on Safe Control Gym, a benchmark designed for evaluating safe control techniques. We show that the proposed method provides a more efficient and optimal solution for high-dimensional control optimization problems, demonstrating significant improvements over existing techniques.