A fusion model combining improved A^* algorithm and dynamic window approach for path planning of obstacle-overcoming robots

as traditional path planning methods often face difficulties and reach excessively long paths when planning paths for mobile robots in complex environments, a novel method that combines the improved $\mathrm{A}^{*}$ algorithm and the dynamic window approach (DWA) is proposed here for path planning of obstacle-overcoming robots. Specifically, point cloud data are obtained by 3D laser radar (LiDAR) scanning and the discrete noises are removed by radius filtering, and the elevation information of each grid is computed to construct raster maps with elevation data; the obstacle region is divided into an obstacle-avoidance area and an obstacle-overcoming area as per the robot’s obstacle-overcoming capacity. Next, based on the improved $\mathrm{A}^{*}$ algorithm and the obstacle-overcoming function, the path return mechanism is introduced to reduce the length of the planned path and number of corner-turnings, thereby optimizing the planned path. Last, the key points in global path planning are set as the target points of the DWA to fuse global and local planning. Experiments reveal that our fusion model has significantly cut the length of the planned path and improved the path-planning flexibility and efficiency of the obstacle-overcoming robot.