A hierarchical-centralized MPC strategy for connected automated vehicular platoon incorporating level interactions

Recently, there has been a surge of interest in developing platoon control strategies based on model predictive control (MPC) to enhance cooperation among connected automated vehicles (CAVs). Nevertheless, solving the optimization problem in MPC instantaneously while achieving centralized coordination among all vehicles proves to be challenging. To address this problem, this study proposes a hierarchical-centralized MPC (HCMPC) strategy for CAVs in a platoon, where a two-level information interaction and control decision generation procedure is developed. In this strategy, interactions among CAVs are divided into two levels by decoupling the platoon into sub-platoons. The upper level denotes the interaction between the leading vehicle of the whole platoon and sub-leading vehicles. The lower level denotes the interaction between the sub-leading vehicle and following vehicles in the same sub-platoon. Corresponding to the two levels, the platoon controller and sub-platoon controllers are interacted to realize cooperative behaviour while reducing computational time. A thorough stability analysis including asymptotic stability and string stability is conducted, obtaining sufficient conditions for different levels of string stability of this novel hierarchical control structure. The results of both numerical and field experiments show that HCMPC reduces the computational time significantly while achieving similar performance to idealized MPC in terms of realizing control target and suppressing traffic oscillations.