Path planning of mobile robots based on improved RRT algorithm

Abstract

Abstract Aiming at the shortcomings of RRT (Rapidly-exploring Research Tree) algorithm such as long time cost and low utilization of sampling points, an improved RRT algorithm is proposed. By adopting the sampling strategy based on dynamic probability, the robot is prevented from falling into a local minimum during the sampling process. At the same time, the adopt of variable step strategy as the random tree expands reduces the number of sampling points. Finally, the initial path is optimized to make it more suitable for the robot to walk. The improved RRT algorithm is compared with the RRT algorithm and the Goal-bias RRT algorithm in both simulations on MATLB and experiments on robot based on ROS (Robot Operating System). And the results show that improved RRT algorithm increases the speed of path planning, reduces the length of path, and the planned path is smoother and more suitable for the real robot to move.