posted on 2015-03-19, 18:22authored byRonan Doolan, Gabriel-Miro Muntean
Increasing amounts of time is wasted due to traffic congestion in both developed and developing countries. This has severe negative effects, including drivers stress due to increased time pressure, reduced usage efficiency of trucks and other commercial vehicles, and increased gas emissions--responsible for climate change and air pollution affecting population health in densely populated areas. As existing centralized approaches were neither efficient, nor scalable, there is a need for alternative approaches. Social insects provide many solutions for dealing with decentralized problems. For instance ants choose their routes based on pheromones left by previous ants. However, Ant Colony Optimization is not directly applicable to vehicle routing, as routing the vehicles to the same road would cause traffic congestion. Yet, the traffic is broadly similar from work- to work-day. This paper introduces an ant-colony optimization-based algorithm called Time-Ants. Time-Ants considers that an amount of “pheromone” or a traffic rating is assigned to each road at any given time in the day. Using an innovative algorithm the vehicle’s routes are chosen based on these traffic ratings, aggregated in time. After several iterations this results in a global optimum for the traffic system. Bottlenecks are identified and avoided by machine learning. Time-Ants outperforms another leading algorithm by up to 19% in terms of percentage of vehicles to reach the destination within a given time-frame.