Research on Grasping of Unknown Object Imitating Human Grasping Reflex for Anthropomorphic

针对拟人机械手提出了一种模仿人类抓取反射的未知物体抓取策略。抓取反射有两个动作：一个是确保积极抓取的抓取动作，另一个是阻止物体释放的保持动作。在建议的抓取中，拇指和手指的每个关节都通过影响其相邻指尖侧链的接触力独立控制。通过设置合适的接触力，可以实现指尖抓取和包络抓取，抓取力均匀。给出了一种具有分布式触觉传感器的拟人机械手Gifu Hand抓取三维未知物体的实验结果。结果表明，具有抓取反射的机械手能够稳定地抓取几何未知物体。

This research presents a grasping strategy of unknown objects imitating human grasping reflex for anthropomorphic robot hands. The grasping reflex has two actions : grasping action which ensures positive grasping, and keeping action which prevents the release of the object. In the proposed grasping, each joint of the thumb and the fingers is controlled independently using the contact force affecting its adjacent fingertip side link. By setting a suitable contact force, both fingertip grasping and enveloped grasping with uniform grasping force are executable. Experimental results of three-dimensional unknown object grasping using an anthropomorphic robot hand, named Gifu Hand, with a distributed tactile sensor, has been presented. It has been shown that the robot hand equipped with the grasping reflex is able to grasp geometrically unknown objects stably.