Disturbance-Robust Blending Control for Gradient-Free Backup Control Barrier Functions: Safety Guarantees for Underpowered Systems in the Presence of Uncertainty

Rapid advances in autonomous systems have led to their widespread integration into modern infrastructure, with increased performance responsibility in complex environments. This necessitates real-time assured systems that are provably robust, reliable, and safe in a variety of known and unknown environments–after all, the methods to guarantee safety are only as good as their ease of implementation. This paper explores methods for provably safe and real-time assured collision avoidance among uncertainties in both state and environments. A robust blending mechanism is employed within the context of gradient-free barrier-function-based safety to guarantee both speed of implementation and safety in dynamic, uncertain environments. The proposed approach is validated both theoretically and empirically, showing robustness to external disturbances where existing methods fail.