Exploring Factors Impacting on the Lane Choice of Riders of Non-Motorized Vehicles at Exit Legs of Signalized At-Grade Intersections

For most signalized at-grade intersections, exclusive lanes for non-motorized vehicles have been applied to improve the level of service, capacity and safety of both motorized vehicles and non-motorized vehicles. However, because of various factors, riders of non-motorized vehicles have been observed using lanes for motorized vehicles instead of lanes for non-motorized vehicles, which usually negatively influences the performance of signalized intersections and sometimes may cause serious problems such as traffic congestion and accidents. The objective of this paper is to explore factors influencing the lane choice of riders of non-motorized vehicles at exit legs of signalized at-grade intersections and develop a prediction model for riders’ lane choice. Data concerning the lane choice of riders of non-motorized vehicles and other impacting factors were collected at exit legs of four typical signalized at-grade intersections. Applying binary logistic regression, a probability prediction model was developed to explain how various factors influence the lane choice of riders of non-motorized vehicles. The prediction model indicates that female riders of non-motorized vehicles have a higher probability of choosing the lane for non-motorized vehicles than male riders. Compared with riders of non-motorized vehicles powered by electricity, riders of traditional man-powered bicycles are more likely to choose the lane for non-motorized vehicles. Right-turning riders of non-motorized vehicles are more likely to choose the lane for non-motorized vehicles than straight-going riders, who in turn, are more likely to choose the lane for non-motorized vehicles than left-turning riders. Decreasing the volume of non-motorized vehicles, increasing the volume of motorized vehicles, and widening the lane for non-motorized vehicles will increase the probability of the correct choice of lane for non-motorized vehicles. The predictions of the model are in good agreement with the observed facts. The model is meaningful for guidance on the design and management of signalized at-grade intersections