Dynamic Path Planning Based on Improved D* Algorithms of Gaode Map

Abstract

To solve the problem of long-distance path planning for outdoor robots, an improved $D^{\ast}$ algorithm combing with Gaode map based on vector model is proposed. Specifically, the global static path to the target point is planned through the Gaode Map Open Platform, and is divided into path sub-nodes. The dynamic $D^{\ast}$ algorithm of heuristic function h(n) is improved under the vector model, and the long-distance path planning is realized by node iterations. The simulation is carried out on Gazebo and Rviz platforms. Results show that compared with the traditional $D^{\ast}$ algorithm, the running time of the robot is reduced by 32.8%, the number of corners is reduced by 64.6%, the number of dead zones is reduced by 64.3%, and the success rate of the planned path to the target point is greatly improved, which has high feasibility.