Online Generation of Reachable Grasps for Dexterous Manipulation Using a Representation of the Reachable Workspace

Abstract

In service robotic tasks, the ability to grasp and handle objects is mandatory. Short response times with respect to execution of commanded tasks are necessary. Planning in general and grasp planning in particular should happen online. We extend the online grasp planner by Borst et al. [1] to generate reachable grasps while preserving the integrity and modularity of the grasp planner. To achieve this a representation of the reachable space of a robot arm is used to determine a grasp’s reachability. Furthermore we show the in uence of obstacles on the reachability throughout the workspace. A method to include obstacles into the representation of reachability is sketched. The resulting representation is used by the grasp planner. The performance of the algorithms is evaluated by measuring their computation times. Even in the worst case our grasp planner outperforms comparable state of the art approaches.