Trajectory Planning Based on MINVO Basis for Autonomous Vehicles in Lane Change Scenarios

This paper focuses on the problem of collision-free lane change for autonomous vehicles under dynamic obstacle road scenarios and proposes a new trajectory planning algorithm for autonomous vehicles. With the goals of dynamics feasibility, comfort, and safety in mind, we develop a B-spline-based trajectory planning method capable of generating trajectories which satisfy all the desired objectives. Specifically, (i) the maximum velocity, the maximum acceleration and the maximum curvature of the trajectory are taken into account explicitly to ensure the comfort of the ride and the traceability of the planned path; (ii) MINVO basis is used to obtain outer polygon representations with minimum area of each interval of the host vehicles' trajectories. These polygon representations are separated by designed artificial hyperplanes to ensure collision avoidance; and (iii) the objective function is designed to guarantee both feasibility and comfort with the given trajectory by incorporating a jerk term and a deviation from the goal point for the penalty. Finally, simulation experiments on both the ample and aggressive scenarios are conducted to verify the effectiveness of this proposed trajectory planning algorithm.