#machine learning
Apr 2026
Safe learning-based control via function-based uncertainty quantification
The authors' uncertainty tubes rely solely on sampled realizations and can therefore accommodate discontinuities represented by the sampling model and are integrated into a safe Bayesian optimization algorithm with which to safely tune control parameters on a real Furuta pendulum.
Abdullah Tokmak, Toni Karvonen, Thomas Bo Schön et al.
· arXiv.org · 1 citation