Parallel cartesian planning in dynamic environments using constrained trajectory planning

Abstract

We present a parallel Cartesian planning algorithm for redundant robot arms and manipulators. We pre-compute a roadmap, that takes into account static obstacles in the environment as well as singular configurations. At runtime, multiple paths in this roadmap are computed as initial trajectories for an optimization-based planner that tends to satisfy various constraints corresponding to demands on the trajectory, including end-effector constraints, collision-free, and non-singular. We highlight and compare the performance of our parallel planner using 7-DOF arms with other planning algorithms. To the best of our knowledge, this is the first approach that can compute smooth and collision-free trajectories in complex environments with dynamic obstacles.