Intelligent Traffic Lights: Green Time Period Negotiaton

Abstract

Abstract: This paper presents a person-based traffic signal control strategy for isolated intersection, including simultaneously signal plan design and signal timing optimization with real-time information on the network dynamics. In order to update the traffic signal control plan, a model of efficient allocation of an available resource (green time period) to consumers (traffic lights) was devised as a protocol to decide who gets the right of using the resources based on an auction-like schema. Traffic streams participate in such an auction to get the right of consuming a certain amount of the available resource and the intersection-control system mediates between traffic streams with opposing goals (conflicting movements). Whereas the decision is to terminate the actual green time, the green time can be assigned to any phase of the traffic signal plan. The selected phase is the one that gives the most benefic contribution for the intersection performance. Decision is formulated to minimize the total person delay at the intersection and consequently assign priority to vehicles with more passenger occupancy, also taking into account the pedestrians. In a social management perspective it should be more important and valuable to minimize “people's” delays or other person-based measure instead of vehicle-centric values. A real four-arm intersection with a time varying origin-destination demand example is tested to demonstrate the proposed method. The paper contributes to the development of a new traffic signal control strategy which breaks with the traditional concepts of traffic control such as the cycle length, the maximum green period and fixed phase sequence. Results reveal that the proposed signal control system reduces total person delay at the intersection and effectively provides priority to vehicles with more occupancy. The system structure is flexible and able to adapt traffic control decisions to predictions and react to unexpected traffic events.