An efficient grasp planning algorithm for 3-D objects considering hand configuration

Precision grasp planning on three-dimensional (3-D) objects has received a lot of attention in the past few years. However, traditional contact-level methods are not applicable for real grasp tasks since they don't consider hand configuration, and heuristic approaches require a large amount of computing time. In this paper, the grasp planning problem is interpreted geometrically and the workspaces of different robotic fingers are described using convex sets. Then an efficient grasp planning algorithm is proposed for arbitrary objects considering hand configuration based on the concept of Q ⁺ distance. The proposed method is suitable for fully-actuated or underactuated multifingered robotic hands. Finally, the algorithm is implemented on a four-fingered robotic hand, and simulation results indicate that the algorithm can search for joint angles and relative pose between the robotic hand frame and the object frame for a valid grasp configuration within one second.