Driver-centric data-driven robust model predictive control for mixed vehicular platoon

The penetration rate of automated vehicles (AVs) may remain unsaturated for a long time, resulting in the coexistence of AVs and human-driven vehicles (HDVs). The non-ideal driving behavior has inherent uncertainties and randomness that would cause traffic congestion and oscillation. To improve the driving safety and comfort of mixed vehicular platoon (MVP) consisting of HDVs and AVs, this paper proposes a driver-centric data-driven robust model predictive control (DDRMPC) strategy. This strategy involves two new elements of a data-driven robust model predictive controller and the personalized driving policy. A data-driven MVP model is established with subspace identification to alleviate the adverse effects of uncertain dynamics. To provide a driver-centric driving experience, a personalized driving policy with a flexible spacing corridor and soft constraints is designed in terms of different driving styles. In this way, a tube-based robust model predictive controller is integrated with the data-driven MVP model to develop the DDRMPC strategy, and its stability is proven. Moreover, four experiments with thirty-seven drivers are carried out on a self-developed MVP platform. Finally, several experiments with MVP demonstrate the effectiveness of the proposed DDRMPC strategy.