A modeling method for two-dimensional two-wheeler driving behavior during severe conflict interaction at intersections
Zhenyuan Liu,Naiting Zhong,Junyi Chen,Bingzhao Gao
DOI: https://doi.org/10.1016/j.aap.2024.107668
Abstract:The safety of two-wheelers is a serious public safety issue nowadays. Two-wheelers usually have severe conflict interaction with vehicles at intersections, such as running red lights, which is very likely to cause traffic accidents. Therefore, a model of two-wheeler driving behavior in conflicting interactions can provide guidance for traffic safety management on one hand, and can be used for the development and testing of autonomous vehicles on the other. However, the existing models perform poorly when interacting with vehicles. To address the problems, this paper proposes a modeling method (an improved social force model, ISFM) for two-dimensional two-wheeler driving simulation for conflict interaction at intersections. Based on analysis of naturalistic driving study data, when two-wheelers encounter with vehicles, their driving intentions and trajectories can be categorized into two groups, which are yielding and overtaking. Therefore, the vehicle-related social forces are designed to be a set of two forces rather than a repulsion force in original SFM, which is a yielding force based on the relative distance between the two-wheeler and the vehicle, and an overtaking force based on the velocity of the two-wheeler itself. This opens up the possibilities for modeling the multi-modal driving intention of two-wheelers encountering with cross traffic. Based on ISFM, a bicycle model, a powered two-wheeler (PTW) model and a model of a group of PTWs, are then constructed. Compared to the original SFM, ISFM increases the precision of driving intention prediction by 19.7 % (yielding situation) and 25.0 % (overtaking situation), and reduces the root mean square error between simulated and actual trajectories by 7.8 % and 14.8 % on the bicycle model and the PTW model, respectively. Meanwhile, the model of a group of PTWs also performs well. Finally, the results of ablation experiments also validate the effectiveness of the social force designed based on velocity.