Incorporating Looking-Back Behavior into Cellular Automaton Model

Shoufeng Ma,Jing Jin,Bin Ran,Shiquan Zhong,Liang Zheng

2012-01-01

Abstract:Traditional cellular automaton (CA) models assume that drivers only receive information from the preceding vehicles, e.g. the brake light information from downstream. However, in reality, drivers not only perceive information from downstream but can also get upstream information by routinely checking traffic conditions through the rearview mirrors. Meanwhile, with the introduction of Connected Vehicle Technologies, the perception range of drivers is expected to significantly increase which can lead to more informed driving behavior with both downstream and upstream traffic conditions. Such impact cannot be easily modeled by traditional forward-looking CA models. In this study, the perception ranges of both looking forward and backward are introduced into the CA model. Driveri¯s behavior change is then formulated when vehicles appear within those ranges. Moreover, the frequency or probability of such i°looking-backi± behavior can be affected by the prevailing traffic condition, i.e. the greater local vehicle density leads to higher tendency for drivers to look back to maintain comfortable driving operations. The critical parameters in the modified CA model are firstly calibrated with the empirical data to create a baseline model. Then, by tuning the model parameters in the baseline model this paper evaluates the impact of the looking-back probability and perception range on the characteristics and stability of traffic flow through numerical experiments. It is found that higher looking-back probability can produce higher flux and benefits the stability of traffic flow. Greater perception range can also contribute to the stability of traffic flow, e.g. the disappearance of S (Synchronized flow)->J (Wide moving jam) phase transition, and the reduction of the duration and vehicle cluster length of traffic jam although it may cause smaller flux.