DADEE: Well-calibrated uncertainty quantification in neural networks for barriers-based robot safety

TL;DR

This paper introduces DADEE, a novel approach combining model variance and direct estimation algorithms to improve uncertainty quantification in neural network-based control barrier functions, enhancing robot safety.

Contribution

It proposes a new method that combines different uncertainty estimation techniques to achieve well-calibrated uncertainty in CBF-based controllers.

Findings

Combined algorithms improve safety in robot control.

Model variance methods excel out-of-domain uncertainty estimation.

Direct estimation methods perform better in-domain.

Abstract

Uncertainty-aware controllers that guarantee safety are critical for safety critical applications. Among such controllers, Control Barrier Functions (CBFs) based approaches are popular because they are fast, yet safe. However, most such works depend on Gaussian Processes (GPs) or MC-Dropout for learning and uncertainty estimation, and both approaches come with drawbacks: GPs are non-parametric methods that are slow, while MC-Dropout does not capture aleatoric uncertainty. On the other hand, modern Bayesian learning algorithms have shown promise in uncertainty quantification. The application of modern Bayesian learning methods to CBF-based controllers has not yet been studied. We aim to fill this gap by surveying uncertainty quantification algorithms and evaluating them on CBF-based safe controllers. We find that model variance-based algorithms (for example, Deep ensembles, MC-dropout, etc.) and direct estimation-based algorithms (such as DEUP) have complementary strengths. Algorithms in the former category can only estimate uncertainty accurately out-of-domain, while those in the latter category can only do so in-domain. We combine the two approaches to obtain more accurate uncertainty estimates both in- and out-of-domain. As measured by the failure rate of a simulated robot, this results in a safer CBF-based robot controller.