Optimum grasp of planar and revolute objects with gripper geometry constraints

This work presents a new methodology for synthesizing force-closure two-finger and three-finger grasps for planar and revolute parts. Most real world objects can be modeled using such parts. The grasp synthesis explicitly takes into account the gripper geometry constraints, but unlike previous related work, we optimize several grasp quality criteria simultaneously, rather than individually. The methodology also has two distinct features that makes it complete: corners are considered to ensure every existing grasp is found and an accessibility check is made around the part contour to avoid synthesizing impossible grasps. The algorithm is therefore practical while it ensures high quality grasps. Results are presented for grasp synthesis of several objects with SARAH, an under-actuated three-finger robotic hand.