Workspace of Multi-fingered Robotic Hands Using Monte Carlo Method

Multi-fingered hands enable significantly enhanced manipulation capabilities to the robot where it is attached. As a consequence, analysis, design and development of multi-fingered hands has been of continuing interest in the robotics community. In this work, we propose a probabilistic Monte Carlo based approach to obtain the workspace of a well-known multi-fingered hand, the three-fingered Salisbury hand, modeled as a hybrid parallel manipulator. It is shown that Monte Carlo method can be used to obtain the volume of the well conditioned workspace of the hybrid manipulator in R-3 and SO(3). One of the obtained novel results is that with realistic constraints on the motion of the joints, the well-conditioned workspace of the hybrid manipulator is the largest when the grasped object area is approximately equal to the palm area. We also obtain and discuss the dependence of the workspace of the manipulator on it's geometry and other link and joint variables.