Asymptotically Time-Optimal Smooth Trajectory Planning in Dynamic Environments
Hang Zhou, Tao Meng, Shujian Sun · 2024
This paper proposed an algorithm for smooth trajectory generation in complex environments with dynamic obstacles and velocity constraints. The proposed two-step algorithm: Tube Space-Time Rapidly-exploring Random Trees Star (TubeSTRRT*), is combined with the improved reformulation dynamic coordinate minimum snap (RDCMS) to generate smooth, collision-free trajectories with asymptotic time optimality. First, the space-time state space is sampled to obtain time information for each node, facilitating the avoidance of moving obstacles. Then, to address the issue of non-smooth paths in TubeSTRRT*, a dynamic tube is generated for each node and combined with the RDCMS to create a smooth, collision-free trajectory. Finally, comparative simulations demonstrate the superiority of our algorithm in terms of smoothness.