| Summary: | Generating a collision-free and dynamically feasible trajectory with a better clearance in a cluttered environment is still a challenge. We propose two dynamically feasible B-spline based rapidly exploring random tree (RRT) approaches, which are named DB-RRT and FMDB-RRT, to achieve path planning and trajectory planning simultaneously for omnidirectional mobile robots. DB-RRT combines the convex hull property of the B-spline and RRT’s rapid expansion capability to generate a safe and dynamically feasible trajectory. Firstly, we analyze the tree’s sustainable growth ability and put forward the dynamically feasible region. A geometric method is proposed to judge whether finding a dynamically feasible trajectory quickly. Secondly, we design two steer functions to guide the tree’s growth, improve efficiency, and decrease the number of iterations. To further increase the clearance and reduce the randomness of the trajectory, we propose FMDB-RRT, which uses the path of fast marching to guide the rapid growth of DB-RRT. Then, assuming that the number of sampled points is sufficient to represent the dynamically feasible region, the DB-RRT is proved to be probabilistically complete. Finally, by conducting experimental comparisons with other algorithms in different environments and deploying the proposed algorithm to an omnidirectional mobile robot, the effectiveness and good performance of the algorithm have been verified.
|
|---|
