MPC-Based Method for Intersection Control in Mixed Traffic Environments with Autonomous Vehicles, Human-Driven Vehicles, and Pedestrians

Abstract

Autonomous vehicles (AVs) will inevitably share the road with human-driven vehicles (HDVs) and pedestrians for a long time in the near future. The paper introduces an intersection management method that comprehensively considers AVs, HDVs, and pedestrians. First, the right-of-way among AVs, HDVs, and pedestrians is assigned by a maximum pressure control-based method based on the queue length estimation for pedestrians and AV–HDV mix flow. Then, to entirely eliminate conflicts between HDVs and AVs, a behavioral decision strategy is further presented for AVs when encountering HDVs within the intersection. Finally, simulation experiments are conducted to validate the model by SUMO platform with Python scripts realizing the proposed method. Results show that the proposed model stabilizes gradually while bounding queue lengths of mixed traffic flows and pedestrians. As the penetration rate of AVs increases, it improves managing intersection networks containing all three types of traffic modes, even when there are unknown penetration rates or turn rates present. The proposed model adapts to changes in demand from vehicles or pedestrians; however, there is a relatively small correlation between vehicle demand and pedestrian queue length.