Simulation of urban non-motorized traffic: A agent-based modeling approach based on big data of bike sharing and social force model

Abstract

Encouraging cycling, one of the urban non-motorized transport modes, has been well recognized as an environment-friendly way that alleviates urban traffic congestions and solves the first/last mile issue. However, concerns about the efficiency and safety of urban cycling have been widely made by urban planners due to the unclear right-of-way and discontinuous non-motorized corridor for cycling. This study uses the dynamic location data of the Meituan Bike (formerly Mobike) in the Hi-tech Park area of Shenzhen to analyze the spatial-temporal variations of bikeshare use, aiming at identifying the traffic corridors of cycling. Combined with agent-based modeling technique and social force model, this research proposes a new approach of simulating the urban non-motorized traffic, and hence provides valuable insights for building bicycle lanes for cycling corridors. The results show that (1) the usage of bike sharing during weekdays is 2.5 times that on weekdays, and the cycling corridors are usually the main and secondary roads that are in the vicinity of the metro stations; (2) adding bicycle lanes can reduce the traffic density of the non-motorized volume by 6% in an overall, and save the travel time of cyclists and pedestrians by 6.4% and 3.7%, respectively.