Abstract Summary
The technological advances in connected and automated vehicles (CAVs) enable cooperative (automated) vehicles to exchange information with not only infrastructure but also among vehicles. Joint control of traffic signals and vehicle trajectories has the potential to improve traffic operations and environmental economy on urban roads. Exiting literature on CAV platooning on urban roads mainly focus on driver assistant systems, cooperative vehicle intersection control algorithms, CAV trajectory optimization, and the integrated optimization of traffic signals and vehicle trajectories. However, most of these algorithms were designed to optimize simple objective functions of the platoon leaders, which cannot reflect the benefits of the whole platoon. We propose a hierarchical approach for joint design of signal timing and cooperative (automated) vehicle trajectories at typical four-arm intersections with predetermined phase sequence. The upper layer of the proposed control approach determines the optimal lengths of signal phases based on optimal control and a surrogate model of CAV platoon dynamics. For simplification, only the accelerations of the platoon leaders and the first-stopping vehicles are optimized, while the other following vehicles are represented using a car-following model. The running cost considering multiple terms (i.e. throughput, comfort, travel delay, safety, fuel consumption) is piecewise according to the signal indication and the vehicle sequence in the platoon. Additional penalty terms for first-stopping vehicles have an advantage on longer horizon like consecutive signal cycles, rather than limited within one signal cycle. The acceleration and speed are constrained within maximal and minimal bounds. The lower layer optimizes trajectories of all vehicles using model predictive control approach under the fixed but optimal signal plan resulted from the upper layer. The signal indication in the lower layer is transferred to the first-stopping vehicle at the beginning of the current signal cycle, which can operate the stopping vehicles smoother than trajectories without communicating the signal information.
