A market-based approach to accommodate user preferences in reservation-based traffic management

Removing the human driver from the control loop by the use of autonomous vehicles and the integration of these with the traffic management infrastructure is a challenging long-term vision for the field of Intelligent Transportation Systems (ITS). Setting out from a recently proposed urban infrastructure that allows for autonomous vehicles to individually reserve space and time inside an intersection to safely cross them, multiagent approaches have been applied to simulate and to manage both single intersections as well as networks of intersections, achieving significant improvements in the drivers average travel times. However, these approaches do not take full advantage of the potential of vehicle-centric ITS, as they ignore the fact that different drivers may value their travel times quite differently. In this paper we combine two different market-based mechanisms, acting at intersection level and at network level, respectively, to accommodate driver preferences in reservationbased urban traffic management. At intersection level, intersection manager agents assign space-time slots through combinatorial auctions, while at network level a pricing scheme, based on general market equilibrium, accounts for an efficient use of the available network resources. Our experiments show that this combined approach on the one hand allows drivers to effectively improve their travel times if they are willing to pay more money for their trip, while on the other hand the negative impact on social welfare (average travel times) is unnoticeable.