An Integer Programming Formulation for Heterogeneous Traffic Dynamics and Assignment Modeling: A Multicommodity Network Flow Model in Space-Time Networks

With the connected automated vehicle (CAV) growing, their interactions with the human-driven vehicles will significantly change the traffic flow characteristics. While traditional traffic flow studies mostly assume molecular driving behaviors, it increasingly becomes arguable for heterogeneous traffic flow because of the different behaviors between CAVs and human-driven vehicles. To address this issue, we present an integer programming formulation for vehicle-oriented traffic dynamics model in space-time networks. In this model, individual vehicles and their driving behaviors can be explicitly distinguished. As such the diverse attributes of heterogeneous traffic flow can be well convoyed. Through analyzing queuing propagations at various bottlenecks, we demonstrate how the proposed model can be used for traffic dynamics analysis. The propagation of backward shockwave for scheduled CAVs can be precisely estimated based on each queuing vehicle's delay, freeflow travel time and travel speed in the queue. Overall, the proposed model can possibly serve as a building block for controlling both queue evolution and traffic assignment for a large-scale heterogeneous traffic at a network level.