Optimal lane-changing path planning for collision avoidance in emergency evasion scenarios

Path planning for lane changes in emergency avoidance scenarios is a multi-objective, multi-constraint optimization problem applicable across various scenarios; Research in this area is one of the key technologies for human-machine collaboration and autonomous driving control. To address the challenges posed by the coupled effects of safety, comfort, and efficiency in multi-objective lane-changing path planning for emergency avoidance, as well as the difficulty in dynamically adjusting the weights of performance objectives, this paper proposes an optimal collision-avoidance lane-changing path planning method that integrates natural driving data features with adaptive multi-criteria evaluation. By utilizing natural driving data from emergency avoidance scenarios, this study quantifies the duration and longitudinal distance of lane changes during actual driving. It then employs a fifth-degree polynomial to generate clusters of feasible lane-changing trajectories and uses the TOPSIS-entropy weighting method to perform a comprehensive evaluation with objective weighting of multi-criteria data, thereby selecting the optimal collision-avoidance lane-changing path. Simulation scenarios are set up and compared. The results indicate that the proposed collision-avoidance optimal lane-changing path planning model can effectively avoid obstacles while achieving adaptive matching of multi-objective weights—including safety, comfort, and efficiency—demonstrating the rationality and effectiveness of this method.