Performance Evaluation of Particle Swarm Optimisation Control for Traffic Light Systems in Roundabouts

Although the usage of roundabouts has been widely implemented as an alternative to signalised intersections as a version of traffic control, it is still reported that major traffic congestions commonly occur at roundabouts in urban areas. However, studies show that one of the most promising methods to overcome this issue is through the implementation of an evolutionary algorithm as a method to control the phase timings of the traffic light system at roundabouts. As such, this project aims to evaluate the effectiveness of implementing particle swarm optimisation (PSO) as the traffic light control for a signalised roundabout. Hence the performance of the PSO-controlled signalised roundabout is compared with that of a fixed-time signalised roundabout and also the unsignalised roundabout under different traffic conditions. This project presents a case study inspired by a real-world roundabout, namely the Persiaran Anggerik Mokara Roundabout in Kota Kemuning, Shah Alam, Malaysia. From the study, we can conclude that the usage of a Particle Swarm optimisation (PSO) algorithm is suitable in finding more effective green traffic light phase timings compared to fixed-time traffic lights for all test cases. However, the implementation of the PSO-controlled signalised roundabout is less effective compared to unsignalised roundabouts due the minimal delay incurred from an unsignalised round design. As such, it was concluded that when designing a roundabout system that is of similar characteristics as the modelled roundabout for this project, an unsignalised approach to the roundabout system greatly outperforms any traffic light system that only has traffic signals at the entrances of the roundabout.