Safe Set

Safe set research focuses on developing methods to guarantee the safety of autonomous systems, particularly in complex scenarios like robotics and reinforcement learning, by defining and maintaining a region of the system's state space where safe operation is ensured. Current research emphasizes data-driven approaches, leveraging control barrier functions (CBFs) and model predictive control (MPC) to synthesize safe sets, often incorporating techniques like sum-of-squares programming for efficient computation. This work is crucial for enabling the safe deployment of advanced autonomous systems in real-world applications, addressing limitations in existing methods that lack provable safety guarantees or struggle with high-dimensional systems and complex constraints.