Certain Types of Vehicles in Heterogeneous Traffic in India Tend to Gather

Abstract

For heterogeneous traffic with weak lane discipline (2D mixed traffic), a quantitative investigation of the arrangement of vehicle types is necessary to understand and accurately simulate traffic phenomena. This study accordingly directly detected the vehicle groups observed frequently in 2D mixed traffic footage collected in Mumbai, India to investigate vehicle herding behavior via leader–follower networks. These networks were extracted as frequent sub-networks in standardized traffic (FSSTs) that assume standardized numbers of vehicles and network structures. The results showed that specific vehicle types tend to herd, FSSTs comprising unique vehicle types or certain sets of vehicle types were predominant, and FSSTs exhibited mutual exclusion of motorcycles and auto-rickshaws. An analysis of FSSTs with unique vehicle types clarified the difference in formations according to vehicle type. These phenomena were discussed based on the segregation and formation generation of pedestrian groups in previous research.