Safe learning-based control via function-based uncertainty quantification
2026-04-01 • Machine Learning
Machine Learning
AI summaryⓘ
The authors focus on safely controlling systems when the rules or behaviors are uncertain. Instead of assuming strict boundaries or smoothness in these unknown parts, they treat the unknown as something random they can sample repeatedly. They create 'uncertainty tubes' that likely contain the true behavior based only on these samples. The authors then use this idea to safely optimize control settings on a Furuta pendulum, showing their method works in practice.
uncertainty quantificationlearning-based controlscenario approachrandom functionsBayesian optimizationsafe controlFuruta pendulumuncertainty tubesi.i.d. samplessafe parameter tuning
Authors
Abdullah Tokmak, Toni Karvonen, Thomas B. Schön, Dominik Baumann
Abstract
Uncertainty quantification is essential when deploying learning-based control methods in safety-critical systems. This is commonly realized by constructing uncertainty tubes that enclose the unknown function of interest, e.g., the reward and constraint functions or the underlying dynamics model, with high probability. However, existing approaches for uncertainty quantification typically rely on restrictive assumptions on the unknown function, such as known bounds on functional norms or Lipschitz constants, and struggle with discontinuities. In this paper, we model the unknown function as a random function from which independent and identically distributed realizations can be generated, and construct uncertainty tubes via the scenario approach that hold with high probability and rely solely on the sampled realizations. We integrate these uncertainty tubes into a safe Bayesian optimization algorithm, which we then use to safely tune control parameters on a real Furuta pendulum.