Abstract:

This paper proposes a driving risk assessment method based on trajectory prediction. Firstly， a driver’s risk field （DRF） with Gaussian cross-section characteristics along both sides of the prediction trajectory is established to characterize the uncertainty of the driver’s behavior. Then， taking the risk consequences of the vehicle and the surrounding static and dynamic obstacles in specific states into consideration， the environmental event cost is established， and the quantitative perception risk that adapts to the uncertainty of complex driving scenarios is obtained. Finally， the quantitative perception risk time series in the prediction interval is then fused based on Bayesian theory to realize prediction of potential collision risks in future driving. The real vehicle trajectory and simulation results show that compared with the classic TTC index method， the risk assessment method of DRF based on the integration of interaction information between self-vehicle and surrounding environment in the future can identify the driving risk changes of complex traffic scenarios faster and more accurately， which provides a reference for the study of vehicle collision risk problems in complex scenarios with multiple surrounding vehicles.

Key words: intelligent vehicle, driving risk assessment, trajectory prediction, complex driving scenes, driver’s risk field