Backup Control Barrier Function Synthesis using Sum-of-Squares Reachability

arXiv:2608.27916 2026 Dynamics 1 ideas extracted · analyzed Sep 1, 2026

What the math gives to ML

The paper provides a constructive finite-horizon SOS backward-reachability mechanism for synthesizing a larger backup set and backup controller, rather than prescribing them manually. Its transferable asset is a polynomial certificate: a time-dependent sublevel set is made invariant under a feasible controller and is constrained to enter an already certified terminal-safe set at the horizon. In neural systems, this can become a runtime safety shield for a learned policy or a certified reachable region for a neural ODE or world model, with the SOS certificate determining exactly when the nominal network must yield to a backup controller.

Ideas from this paper

Mechanism confirmed, baseline not beaten 2026

SOS Backup Shield for Learned Policies

Wrap a neural policy with a backup controller synthesized by finite-horizon SOS backward reachability. The neural policy is used whenever it remains inside the certified feasible region; otherwise, a time-indexed backup controller drives the state into a terminal-safe set while respecting actuator limits.

Useful7/10
Difficulty7/10
Novelty7/10
Paper: Backup Control Barrier Function Synthesis using Sum-of-Squares Reachability arXiv:2608.27916