No abstract

Autonomous vehicles offer new traffic behaviors that could revolutionize transportation. Examples include reservation-based intersection control and reduced reaction times that result in greater road capacity. Most studies have used microsimulation models of those new technologies to study their impacts more realistically. However, microsimulation 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 and arterial intersections because the capacity allocations of signals were more optimized for the network. Reservations also were less efficient than were traditional merges and diverges for on- and off-ramps. However, the increased capacity from reduced following headways resulted in significant improvements for both freeway and arterial networks. Finally, the authors 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.

Reservation-based traffic control is a revolutionary intersection management system which involves the communication of autonomous vehicles and an intersection to request space-time trajectories through the intersection. Although previous studies have found congestion and throughput benefits of reservation-based control that surpass signalized control, other studies have found negative impacts at peak travel times. The main purpose of this paper is to find and characterize favorable mixed configurations of reservation-based controls and signalized controls in a large city network which minimize total system travel times. As this optimization problem is bi-level and challenging, three different methods are proposed to heuristically find effective mixed configurations. The first method is an intersection ranking method that uses simulation to assign a score to each intersection in a network based on localized potential benefit to system travel time under reservation control and then ranks all intersections accordingly. The second is another ranking method; however, it uses linear regression to predict an intersection’s localized score. Finally, a genetic algorithm is presented that iteratively approaches high-performing network configurations yielding minimal system travel times. The methods were tested on the downtown Austin network and configurations found that are less than half controlled by reservation intersections that improve travel times beyond an all-reservation controlled network. Overall, the results show that the genetic algorithm finds the best performing configurations, with the initial score-assigning ranking method performing similarly but much more efficiently. It was finally find that favorable reservation placement is in consecutive chains along highly trafficked corridors.

No abstract