Modeling lane-changing duration of intelligent vehicles by incorporating starting point pre-aiming hypothesis and generalized force method

Existing research on lane-changing duration still faces challenges that require breakthroughs. These challenges include an unclear understanding of the interaction mechanisms during lane-changing, the unknown impact of vehicle type heterogeneity on lane-changing duration, and the lack of quantitative characterization of the differences in lane-changing directions. To this end, a comprehensive study on lane-changing duration modeling was conducted by incorporating relevant mechanisms. Firstly, the lane-changing duration law was mined from human naturalistic driving data, revealing the vehicle type heterogeneity and the directional asymmetry in lane-changing duration. Secondly, the starting point pre-aiming hypothesis was presented in conjunction with the lane-changing trajectory fitting model to illustrate the planning process of lane-changing duration. Finally, generalized force methods were introduced to characterize the influence of surrounding vehicles as a nonlinear mapping of the lane-changing vehicle. Thus, a lane-changing duration model was formed, incorporating the starting point pre-aiming hypothesis and the generalized force method. Comparative validation and sensitivity analysis of the models were conducted. The results demonstrate that the formed model enables quantitative analysis of lane-changing duration from a mechanistic perspective. Specifically, the model incorporating vehicle type heterogeneity and directional asymmetry outperforms models neglecting these factors in prediction accuracy. The study establishes an intuitive and concise model for lane-changing duration and provides a reference for the human-like design of autonomous lane-changing systems for intelligent vehicles.