Estimating Queue at Traffic Signals

Abstract

This paper proposes a new queue prediction model based on the data that can be collected from a single loop detector positioned at the stop line of signalised intersections. A number of different model forms were explored using an enhanced NGSIM dataset. These data were filtered to represent the data that can be typically collected from a stop line detector loop. The best six models resulted in an accuracy ranging from 83% to 95% to correctly predict the state of vehicle’s discharge close to the stop line that is whether it is a queued or platooned vehicle. When combined with a logical filter to group sequential vehicles, it enables a traffic controller to estimate the most likely queue length. The proposed model will form part of a new offset optimizer algorithm currently under development.