Safety in safe Bayesian optimization and its ramifications for control

TL;DR

This paper addresses safety challenges in Bayesian optimization for control parameter tuning, proposing a Lipschitz-based safe BO method that avoids unreliable uncertainty bounds and is applicable to higher-dimensional problems.

Contribution

The authors introduce LoSBO, a safe Bayesian optimization algorithm relying solely on Lipschitz bounds, and LoS-GP-UCB, a variant suitable for higher-dimensional spaces, improving safety and applicability.

Findings

Lipschitz-only Safe Bayesian Optimization (LoSBO) ensures safety without uncertain bounds.

LoS-GP-UCB extends safe BO to higher-dimensional problems.

Numerical experiments demonstrate safety violations in existing methods and effectiveness of proposed algorithms.

Abstract

A recurring and important task in control engineering is parameter tuning under constraints, which conceptually amounts to optimization of a blackbox function accessible only through noisy evaluations. For example, in control practice parameters of a pre-designed controller are often tuned online in feedback with a plant, and only safe parameter values should be tried, avoiding for example instability. Recently, machine learning methods have been deployed for this important problem, in particular, Bayesian optimization (BO). To handle safety constraints, algorithms from safe BO have been utilized, especially SafeOpt-type algorithms, which enjoy considerable popularity in learning-based control, robotics, and adjacent fields. However, we identify two significant obstacles to practical safety. First, SafeOpt-type algorithms rely on quantitative uncertainty bounds, and most implementations replace these by theoretically unsupported heuristics. Second, the theoretically valid uncertainty bounds crucially depend on a quantity - the reproducing kernel Hilbert space norm of the target function - that at present is impossible to reliably bound using established prior engineering knowledge. By careful numerical experiments we show that these issues can indeed cause safety violations. To overcome these problems, we propose Lipschitz-only Safe Bayesian Optimization (LoSBO), a safe BO algorithm that relies only on a known Lipschitz bound for its safety. Furthermore, we propose a variant (LoS-GP-UCB) that avoids gridding of the search space and is therefore applicable even for moderately high-dimensional problems.