Traffic Simulator: A Traffic Risk Hotspot Identification Platform

Qingxian Tang,Changzhen Liu,Jiannan Gao,Liwei Si,Longbiao Chen,Cheng Wang

DOI: https://doi.org/10.1109/swc57546.2023.10448813

2023-01-01

Abstract:With the increment of urban population, the number of vehicles on the road is also constantly increasing, leading to a rise in traffic risks. Frequent urban traffic accidents significantly threaten people’s lives, property, and urban traffic operation. Moreover, traffic violations on the road can cause disorderly traffic, resulting in even more accidents. In order to reduce traffic accidents and ensure people’s safety, urban authorities have to formulate targeted traffic safety strategies for traffic accident risk areas in cities. In this paper, we develop Traffic Simulator, a simulator that can comprehensively, dynamically, and efficiently infer and diagnose traffic risk hotspots in the city. Firstly, we import multi-source heterogeneous data sources from Xiamen, including 3D models, vehicle trajectory data, POIs, and weather into Unity. Secondly, we propose several traffic risk identification algorithms based on the simulator. Thirdly, we perform weighted clustering on identified traffic risk behaviors to obtain traffic risk hotspots. To evaluate the Traffic-Simulator, we conducted extensive research on large-scale and real vehicle GPS trajectories collected in Xiamen. The results show that the framework can effectively and efficiently infer traffic risk hotspot areas in cities.