Lane Change Trajectory Planning for Personalized Driving Comfort and Mobility Efficiency
Date:
Presented our work on personalized lane-change trajectory planning at the 2026 IEEE Intelligent Vehicles Symposium (IV) in Detroit.
This work proposes a neural network-driven planner that integrates a third-order polynomial trajectory generator with an MLP-based learning module for fast optimal lane-change trajectory computation under diverse driving conditions.
The planner uses a shared-backbone dual-head architecture to combine baseline feasibility with personalized comfort- or mobility-oriented behavior. A statistical head-gated switching mechanism based on error-winner logistic regression adaptively selects between the baseline and personalized heads according to the driving context.
Representative cases and Monte Carlo simulations show that the proposed planner can generate individualized lane-change trajectories that trade off driving comfort and mobility efficiency, supporting driver-adaptive planning for autonomous and assisted driving systems.