An algorithm is described which is able to plan collision free motions fast for manipulators with six degrees of freedom (DOF) within 3D environments. After a short review of other implemented 3D motion planners, an outline of the CGR search (combination of goal-directed and randomized search) is given. CGR search pursues a heuristic strategy to effectively find a path in even high-dimensional configuration spaces (C-spaces). The search space, however, is not constructed explicitly, but only tested in certain points determined by the search. Every test whether a point lies in free C-space or not requires a geometric intersection computation. This collision test is very efficient due to a hierarchical object representation. The experimental results show the performance of the proposed motion planner in several realistic tasks, including a snake-like manipulator with six degrees of freedom and a PUMA robot reorienting a bulky load through a gate.
展开▼
