Fast and close to optimal trajectory generation for articulated robots in reaching motions
Khoi Hoang Dinh, Philipp Weiler, Marion Leibold, Dirk Wollherr · 2017
This paper introduces an extension to a method for fast and close to optimal trajectory generation for articulated robots in the case where dynamic constraints and real-time capability are considered. We aim at solving reaching motion problems without pre-defined timing requirements, where the robot starts from its current state trying to reach the final desired state. Our approach combines the advantage of Sequential Action Control (SAC) with an indirect optimization process to further improve optimality and not violate final state constraints while still being applicable for online implementation. Simulation results on a 2 degree of freedom (DOF) and a 3 DOF KUKA-based platform are evaluated to show the improvement of our method in comparison to SAC and other optimal control methods in terms of cost efficiency and computation time.