Unknown Objects Grasp Planning Using Force/Torque Measurements

Geometrically unknown objects grasping is a challenging problem in robotic grasping and most researchers try to find suitable grasp through model reconstruction by vision. In this paper, a new three-fingered robotic hand grasping system is designed for three-dimensional geometrically unknown objects grasping. A six-axis force/torque sensor is mounted on each fingertip in our system to calculate the contact point and corresponding normal direction on the object once contact occurs. Thus it is capable of "exploration" the shape of unknown objects through continual contact to find suitable force-closure grasp configurations. We propose an intuitive algorithm to search possible good grasps and implement it in our system by means of a virtual instrument in LabVIEW (TM). The experimental results validate the feasibility of our grasp system.