Effective Real-Time Urban Traffic Routing: An Automated Planning Approach

Abstract

One of the pivotal challenges presented to urban road traffic controllers is the effective utilisation of transport infrastructure, as a result of growing urbanisation, the finite network capacity, and of the increasing number of road vehicles. In this context, the arrival of connected vehicles present a unique opportunity for a fundamental change in urban traffic control. Urban traffic control approaches should then take an active role in integrating connected vehicles into the mobility ecosystem in order to maximise benefits.To support such integration, in this work we propose to leverage automated planning, a well-studied branch of artificial intelligence, to perform real-time traffic routing in urban areas. We describe the proposed approach, and we demonstrate its effectiveness using real-world historical data of a UK town.

Towards a Framework for Intelligent Urban Traffic Routing

“…Centralised traffic routing techniques in the light of new revolutionary technologies such as CAVs are getting more importance due to their ability to look at the problem of traffic routing from a global perspective which might be important to prevent traffic congestion in some area by smart routing of the vehicles. Recently, automated planning techniques have shown some promise in centralised traffic routing (Chrpa, Vallati, and Parkinson 2019;Vallati and Chrpa 2021b). Such an advancement motivated our work that aims at incorporating centralised routing techniques into the SUMO simulator that provides a realistic simulation of urban traffic (Lopez et al 2018).…”

“…Automated planning approaches had recently gained promising results in centralised traffic routing (Chrpa, Vallati, and Parkinson 2019;Vallati and Chrpa 2021a) as well as in traffic light control (Vallati et al 2016;Pozanco, Fernández, and Borrajo 2021). That inspired us to develop a framework that integrates an automated planning component, responsible for intelligent traffic routing, in the well known SUMO simulator (Lopez et al 2018) in order to evaluate and study its impact in realistic traffic settings.…”

In Defence of Good Old-Fashioned Artificial Intelligence Approaches in Intelligent Transportation Systems

“…Navigating vehicles on segments with higher traffic intensity is penalised and hence the planning engine tries to distribute traffic such that the traffic intensity level on the road segments is as low as possible. A subsequent work involves numeric planning, in which the model is specified in PDDL 2.1 language [37]. A more recent work in the area uses a PDDL+ model that captures the traffic situation more accurately [38].…”