Based on GMM-RRT* Algorithm for Path Planning Picking Kiwifruit Manipulator

Rapidly-exploring random tree (RRT) algorithm exhibits excessive random sampling and poor convergence when planning trajectory in a picking environment. An improved RRT* algorithm employing Gaussian mixture model sampling (GMM-RRT*) algorithm is proposed to optimize the disadvantages of the RRT* algorithm. To verify real effect, GMM-RRT* algorithm, RRT algorithm, RRT* algorithm, RRT* algorithm based on goal node guidance (T-RRT*) algorithm and GMM-RRT* algorithm are worked in 2D scene, 3D scene and manipulator picking simulation scene to solve optimal trajectory, respectively. Results obtained are statistically analyzed in terms of effective length, planning time and number of redundant nodes. Conduct experiments on the Elfin E05 manipulator platform to test the GMM-RRT* algorithm comparing with RRT* algorithm and RRT algorithm. Both simulation and experimental results show that the proposed GMM-RRT* fusion algorithm can effectively shorten planning path time in kiwifruit intelligent picking and improve efficiency of kiwifruit picking.