Probabilistic Safety Constraints for Learned High Relative Degree System Dynamics

TL;DR

This paper introduces a Bayesian learning approach for online system dynamics modeling that ensures safety through probabilistic constraints, enabling autonomous adaptation without offline identification.

Contribution

It presents a novel method combining Bayesian learning with chance-constrained control barrier functions for safe online adaptation of high relative degree system dynamics.

Findings

Successfully maintains safety with high probability during online learning

Enables autonomous model estimation without offline system identification

Demonstrates effectiveness on complex dynamic systems

Abstract

This paper focuses on learning a model of system dynamics online while satisfying safety constraints.Our motivation is to avoid offline system identification or hand-specified dynamics models and allowa system to safely and autonomously estimate and adapt its own model during online operation.Given streaming observations of the system state, we use Bayesian learning to obtain a distributionover the system dynamics. In turn, the distribution is used to optimize the system behavior andensure safety with high probability, by specifying a chance constraint over a control barrier function.