Exploiting automated planning for efficient centralized vehicle routing and mitigating congestion in urban road networks

Abstract

One of the pivotal challenges presented to urban traffic controllers is the effective utilization of the transport infrastructure, as a result of the growing urbanization and of the increasing number of vehicles.In response to this challenge, here we investigate the use of Automated Planning for reducing congestion and maximizing the exploitation of urban networks, by generating routes for the vehicles navigating through the controlled region. Empirical analysis is performed on four case studies (including maps of actual UK urban areas) with different complexities and volume of vehicles. The results demonstrate the potential of the presented approach, leading towards an intelligent navigation system, which shows to be able to provide effective personalized directions to vehicles.

Reducing Traffic Congestion in Urban Areas via Real-Time Re-Routing: A Simulation Study

“…We then considered a second scenario, with heavy traffic conditions. To simulate heavy traffic, we assumed that twice the normal amount of vehicles (3,800) are accessing the network over the same 1 hour simulation time. We are again assuming that all the vehicles are connected and are following the re-routing instructions.…”

“…In consequence, two different vehicles might take different routes even though they have identical locations of origin and destination. Leveraging Automated Planning to tackle the problem of intelligent vehicle routing throughout the road network achieved promising results [3,2]. Those techniques approach the problem from centralised perspective, i.e., there is a central urban traffic controller that suggests routes to all the navigating vehicles [11].…”

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

“…Automated planning techniques can be used to in a centralised fashion to generate a plan for multiple vehicles. Preliminary works leveraging automated planning techniques in such a fashion aim for minimising the risk of road congestion [15] or satisfying air quality constraints [16]. The preliminary results presented in those works indicate that using the "global" planning approach is promising (e.g.…”

“…roads might not become congested even in rush hours), although it might not scale well with an increasing number of considered vehicles and/or the size of the road network. We took inspiration from the work of Chrpa et al [15] in terms of leveraging automated planning for intelligent vehicle routing. However, the approach that we present in this paper can perform realtime traffic routing, rather than the offline approach of Chrpa et al Further, we consider a different planning formalism, namely numerical planning, that provides an excellent tradeoff between computational complexity and accuracy of the knowledge models.…”

Centralised Versus Decentralised Traffic Optimisation of Urban Road Networks: A Simulation Study

“…The work of Chrpa et al [6] exploits AI-based automated planning techniques for enforcing air-quality constraints in urban traffic. In a more recent work, Chrpa et al [7], showed that AI-based planning techniques can be used for reducing congestion of a controlled urban region by re-routing vehicles, but their work is at an exploratory stage and can cope only with a very limited number of vehicles.…”