Safety-Critical Optimal Control for Autonomous Systems

This paper presents an overview of the state art for safety-critical optimal control autonomous systems. Optimal methods are well studied, but become computationally infeasible real-time applications when there multiple hard safety constraints involved. To guarantee such constraints, it has been shown that optimizing quadratic costs while stabilizing affine systems to desired (sets of) states subject and can be reduced a sequence Quadratic Programs (QPs) by using Control Barrier Functions (CBFs) Lyapunov (CLFs). The CBF method is efficient, easily satisfaction nonlinear systems, its wide applicability still faces several challenges. First, with high relative degree, above mentioned QPs if tight or time-varying bounds resulting solution also sub-optimal due myopic solving approach. Finally, this works conditioned on system dynamics being accurately identified. authors discuss recent solutions these issues then present framework combines CBFs, hence termed OCBF, obtain near-optimal guaranteeing even in presence noisy dynamics. An application OCBF approach included vehicles traffic networks.