High-probability Safety Robust Model Predictive Control for Adaptive Cruise Control with State-dependent Uncertainties^*

Adaptive cruise control (ACC) system allows vehicles to autonomously follow one another. However, uncertainties originating from the vehicle model and sensor errors may lead to situations wherein predetermined safety constraints are violated. Robust model predictive control (MPC) methods such as tube MPC adopted to tackle this problem are often criticized for their conservativeness in optimizing for constraint satisfaction. This study proposes an innovative dynamic tube MPC method for ACC to reduce the conservativeness of tube MPC. We prove that the proposed method can guarantee constraint satisfaction with a high probability in the presence of state-dependent uncertainties. Experimental results show that we can get safety constraint sets over the iterations and obtain less conservative controls than classic tube MPC.