Efficient bitmap-based implementation of sequential framework for motion planning for manipulators with many degrees of freedom

Abstract

Our sequential approach is a framework for developing practical motion planners for manipulators with many degrees-of-freedom (DOFs). The essence of this approach is to exploit the serial structure of manipulator arms and decompose the m-dimensional problem of planning collision-free motions for an n-link manipulator into a sequence of smaller m-dimensional subproblems, each of which corresponds to planning the motion of a subgroup of m-1 links along a given path. In this paper, we present an efficient bitmap-based implementation of the sequential framework. This bitmap-based implementation is more efficient and robust than a previously reported visibility graph-based implementation; and it utilizes a novel and more efficient backtracking mechanism. Our implementations are in C running on a SUN Sparc 10. We have conducted extensive experiments for planar arms with up to 8 degrees of freedom among randomly placed obstacles. The experiments show that the bitmap-based implementation of the sequential framework with the novel backtracking mechanism is very efficient. The average run time for a six degree of freedom manipulator in quite cluttered environments are around seven minutes. The planner succeeds for 100% of the examples in our simulations with small backtracking levels (2 for 4 DOF arm, 3 for 6 and 8 DOF arms). >