Transit Demand and Routing after Autonomous Vehicle Availability

Autonomous vehicles offer new traffic behaviors that could revolutionize transportation, such as the reservation-based intersection control and reduced reaction times that result in greater road capacity. Most studies have used micro-simulation models of these new technologies to more realistically study their impacts. However, micro-simulation is not tractable for larger networks. Recent developments in simulating reservation-based controls and multiclass cell transmission models for autonomous vehicles in dynamic traffic assignment have allowed studies of larger networks. This paper presents analyses of several highly congested arterial and freeway networks to quantify how reservations and reduced reaction times affect travel times and congestion. Reservations were observed to improve over signals in most situations. However, signals outperformed reservations in a congested network with several close local road-arterial intersections because the capacity allocations of signals were more optimized for the network. Reservations also were less efficient than traditional merges/diverges for onand off-ramps. On the other hand, the increased capacity due to reduced following headways resulted in significant improvements for both freeway and arterial networks. Finally, we studied a downtown network, including freeway, arterial, and local roads, and found that the combination of reservations and reduced following headways resulted in a 78% reduction in travel time.