Safe and active parameter exploration for event-triggered control

TL;DR

This paper introduces a Gaussian process-based active learning framework to explore parameter spaces in event-triggered control, ensuring safety and convergence through theoretical guarantees and numerical validation.

Contribution

It proposes a novel active learning method for safe parameter exploration in event-triggered control with theoretical safety and convergence guarantees.

Findings

Effective parameter exploration demonstrated in simulations

Theoretical guarantees for safety and convergence

Improved safety and performance in control systems

Abstract

This paper presents a framework of learning parameter space for event-triggered control. In particular, our goal is to find a set of parameters for the event-triggered condition, such that certain specifications on safety and convergence properties are satisfied. The exploration strategy is based on the Gaussian process-based active learning, in which, for each iteration, the parameter with having the largest variance is evaluated. Moreover, we provide a theoretical analysis, so that the derived parameter space satisfies both convergence and safety. Finally, a numerical simulation is given to illustrate the effectiveness of the approach.