Citation

Chen H, Zhang X, Yang W, Jin W, Zhu W, Zhao B. Transportmetrica B: Transp. Dyn. 2021; 9(1): 172-197.

Copyright

(Copyright © 2021, Hong Kong Society for Transportation Studies, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/21680566.2020.1828197

PMID

unavailable

Abstract

In this paper, pedestrian path prediction at a signalized crosswalk with pedestrian-electric bicycle-vehicle mixed flow is investigated. Firstly, a waiting/crossing decision model (W/CDM) is developed to predict pedestrians' waiting/crossing intentions with approaching vehicles. Secondly, a Modified Social Force Model (MSFM) is developed by taking the evasion with conflicting road users, the reaction to traffic lights and crosswalk boundary into consideration. The influence of pedestrians' heterogeneous characteristics and mixed traffic flow are considered for the first time. Then the mixed traffic data at a signalized crosswalk is recorded and analysed. A maximum likelihood estimation is proposed to calibrate the parameters. Finally, the integrated method (W/CDM-MSFM) is compared with the existing methods. The results show that the method outperforms the existing methods and accurately predicts the pedestrians' paths, which makes it possible for autonomous vehicles to present the feasibility to protect the safety of pedestrians and improve the efficiency of dynamic traffic.

Language: en

Keywords

Mixed traffic flow (pedestrian-electric bicycle-vehicle mixed flow); modified social force model (MSFM); pedestrian path prediction; signalized crosswalk; waiting/crossing decision model (W/CDM)