Vehicle Formation Reference Line Smoothing Method Based on Improved FemPosDeviation Algorithm

To address the issues of reference point regression and abnormal jitter in the reference line of following vehicles during specified formation driving of human-driven and unmanned vehicle formation, which make it difficult to keep the formation. This paper proposes an improved FemPosDeviation algorithm for generating a smooth reference line for following vehicles. The method employs a path generation approach for following vehicles based on the Frenet coordinate system and formation geometric parameters to obtain reference path points. To mitigate potential reference point regression in the following vehicle's reference path, an anomaly detection and optimization processing method for the reference line is introduced. Furthermore, to improve the poor consistency of traditional reference line optimization methods, a multi-stage optimization-based reference line smoothing approach is proposed to achieve stable reference line information. Finally, experiments are conducted to validate the feasibility of the method. The results demonstrate that this method effectively resolves the reference line regression issue and, compared to traditional methods, provides better reference line consistency.