Abstract: In view of the problem of path optimization for multi-joint manipulator, an improved RRT (rapidly-exploring random tree) optimization algorithm is proposed. An initial feasible path is searched by using the standard RRT algorithm, then the path cost is calculated on the basis of path length and path security. During the growth of later exploring tree, the middle target node is not randomly sampled, but is selected according to the constraint that it can make the current path cost lower than the former one. Also the selected node should be checked for the distance in order to avoid generating too intensive nodes. For the purpose of speeding up the expansion of the exploring tree toward unknown regions, a greedy heuristic expansion algorithm is used in the process of expanding from the nearest node toward the middle target point, so that the node will expand circularly at a certain step size until it reaches the target node or produces unconnected node. Finally, the path planning simulation experiment is carried out for the 6-DOF maintenance manipulator, the result shows that the planned path quality is greatly improved in comparison to the standard RRT algorithm.

Key words: Greedy heuristic expansion, Manipulator, Path planning, Path security, Rapidly-exploring random trees