Manipulability analysis

Abstract

The ability of computing a quality index for given configurations can be useful for several applications in the context of robotic manipulation. E.g. it can be used for monitoring the current state of the system or it can support decision processes, such as grasp selection in humanoid robotics. Here, a large set of precomputed grasps for a given object have to be quickly filtered in order to select the reachable sub set, for which the inverse kinematics (IK) problem has to be solved. In this work, we present an approach for analyzing the workspace capabilities of a manipulator in order to store a representation for efficient online processing. Compared to existing work, where usually reachability information is used to represent the robot's capabilities, we show how an extended manipulability measure can be used to build a quality distribution in workspace. The proposed approach for manipulability measurement is suitable for redundant manipulators while considering joint limits and self-distance. Further, task specific distributions are introduced and several applications for the humanoid robots ARMAR-III [1] and iCub [2] are presented.