Automatic grasp planning for visual-servo controlled roboticmanipulators

摘要

Vision based grasping holds great promise for grasping in dynamicnenvironments where the object and/or robot are moving. The papernintroduces a grasp planning approach for visual servo controlled robotsnwith a single camera mounted at the end effector. Sensory control,nmechanical, and geometrical issues in the design of such an automaticngrasp planner are discussed and the corresponding constraints arenhighlighted. In particular, the integration of visual feature selectionnconstraints is emphasized. Some quality measures are introduced to ratenthe candidate grasps. The grasp planning strategy and implementationnissues in the development of an automatic grasp planner (AGP) arendescribed. Simulation and experimental results are presented to verifynthe correctness of the approach and its effectiveness in dealing withndynamic situations