Dynamic origin-to-destination routing of wirelessly connected, autonomous vehicles on a congested network

TL;DR

This paper investigates how real-time wireless communication can optimize autonomous vehicle routing on congested networks, comparing simple and complex algorithms through simulations.

Contribution

It introduces and evaluates a dynamic routing approach for autonomous vehicles using real-time data, highlighting the effectiveness of simple congestion-based algorithms.

Findings

Simple congestion-based routing performs as well as more complex methods.

Real-time information significantly improves trip times.

No traffic lights are needed for optimal flow.

Abstract

Up-to-date information wirelessly communicated among vehicles can be used to select the optimal route between a given origin and destination. To elucidate how to make use of such information, simulations are performed for autonomous vehicles traveling on a square lattice of roads. All the possible routes between the origin and the destination (without backtracking) are of the same length. Congestion is the only determinant of delay. At each intersection, right-of-way is given to the closest vehicle. There are no traffic lights. Trip times of a subject vehicle are recorded for various initial conditions using different routing algorithms. Surprisingly, the simplest algorithm, which is based on the total number of vehicles on a route, is as good as one based on computing travel times from the average velocity of vehicles on each road segment.