An adaptive rapidly-exploring random tree algorithm for assembly path planning in complex environments

We present a refined Rapidly-exploring Random Tree (RRT) algorithm for assembly path planning in complex environments. This algorithm adapts its expansion automatically to explore complex environments with narrow passages and cluttered obstacles more efficiently. In this algorithm, the nodes in the tree are classified by various criterions and different extending values are assigned on them indicating the nearby environment and are used to control the future expansion. A series of tree extending schemes are designed and selectively used based on the attributes of the node and the extending result in each step. We show that the algorithm becomes greedy in constrained environments and promising nodes have higher priority to extend than the non-promising ones. The algorithm is evaluated and applied in assembly path planning. The results show significant performance improvement over the standard RRT planner.