A scalable optimizer for automatically generated manipulator motions

Starting from a collision-free robot path-given by an automatic motion planner or by a human programmer-the proposed method shortens and smoothes this path by using iteratively a few local, elementary geometric operations. The fast path-optimization process is interruptable at any time, thus always getting the best result obtainable within a certain amount of time. This property is especially useful in interactive or autonomous applications where the computation time is much more important than the traverse time and the quality of the robot motion. Some example environments including obstacle-constraint tightness and a bulky payload show the performance and usefulness of the new algorithm.<
>