Reachable Set

Reachable set computation aims to determine the set of all possible future states of a system, given initial conditions and constraints. Current research focuses on improving the accuracy and efficiency of reachable set approximations for complex systems, particularly those involving neural networks as controllers or components, employing techniques like Hamilton-Jacobi reachability, zonotopes, Taylor models, and various refinement strategies (temporal and spatial). These advancements are crucial for verifying the safety and reliability of autonomous systems, particularly in safety-critical applications like autonomous driving and robotics, by providing formal guarantees beyond simulation-based testing. The development of efficient and less conservative algorithms remains a key focus.