A Mapping Method of Grasping Posture Applying to the Metamorphic Multi-fingered Hand

As for a tele-operation system of a dexterous robot hand, the core issue is to transform the information of a human grasping posture to the robot hand. Thus, it is essential to obtain a mapping method from a human grasping posture to the metamorphic multi-fingered hand. Joint-to-joint mapping, point-to-point mapping and pose mapping are three commonly used methods for tele-operation. A satisfactory solution could be obtained by using them when we analyze the grasping posture mapping of an anthropomorphic robot hand. But when it comes to the mapping of a non-anthropomorphic robot hand, it is difficult to obtain a satisfactory solution due to structure differences between a human hand and a non-anthropomorphic robot hand. This paper introduces a new mapping method applying to the metamorphic multi-fingered hand considering its special characteristic of the topological structure. The principle of the mapping method is to find a virtual sphere as a carrier in the human frame which could describe the information of the human grasping posture, to transform them to the robot frame, and then to compute the ideal fingertip positions according to the transformed parameters. In the end, an index measuring the mapping result is put forward through comparing differences of ideal central angles and actual central angles of the metamorphic multi-fingered hand.