Personalized Lane Changing Decision-Making Based On Vehicle Environment Interaction

Lane-changing decisions have emerged as a critical challenge in Advanced Driver Assistance Systems (ADAS). Many existing research approaches lack the ability to adapt to individual driver characteristics. Inadequate alignment between decision-making models and a driver's unique driving habits can lead to decisions that are inconsistent with the current traffic environment, thereby compromising traffic safety. To address these limitations, we propose a Personalized Lane Changing Decision (PLCD) model. Our approach first constructs a lane-changing interaction behavior model that incorporates personalized parameters, such as lane-changing duration and driver acceptance gap. Subsequently, The PLCD model is engineered to capture real-time driving conditions and surrounding vehicle data to produce safe and personalized lane-changing decisions. The model's effectiveness is validated through experiments conducted under naturalistic driving conditions. The results show that the PLCD model substantially improves the accuracy of lane-changing decisions, with enhancements of 15.32%, 7.26%, 1.61%, and 10.98% compared to traditional LSTM, SVM, XGB, and KNN models, respectively.