Managing traffic evacuation with multiclass connected and autonomous vehicles

Connected and autonomous vehicles (CAVs) provide a novel perspective to address challenges of traditional evacuation modes, such as the need for trained human drivers, out-of-control, congestion, and limited road capacity. This paper focuses on managing a multiclass traffic evacuation task of private CAVs and mass-transit CAVs. Firstly, we propose a multiclass cell transmission model with moving bottlenecks to model the multiclass CAVs. In particular, we discretize the road network into a multi-size cell network to capture the speed difference between two types of CAVs. The mass-transit CAVs are treated as moving bottlenecks, which can linearly reduce the road capacity in a certain density range. Secondly, we formulate a system optimum collaborative evacuation model to minimize the evacuation network clearance time or minimize the total travel time of evacuees. Constraints include multiclass fleet size, signal-free intersections, loading multiclass CAVs, and non-holding back. Finally, we conduct numerical experiments to test the collaborative evacuation model. On an evacuation corridor, the results show that our proposed model can capture multiclass traffic dynamics and traffic congestion. In the Sioux-Falls network, we evaluate the evacuation efficiency of multiclass CAVs using the fully mixed approach and the lane-based approach. The results indicate that the evacuation efficiency of using the fully mixed approach may be better than that of using the lane-based approach under certain evacuation demands. The cooperation of multiclass CAVs can transfer congestion and reduce evacuation time.