Abstract: Real-time grasping force optimization is a difficult problem because the friction forces between dextrous robot fingers and an object are nonlinear constrains. In this paper, we present a general solution for the optimization problem of multi-fingered grasping force. The method is based on the Lagrange multipliers. The initial grasping forces that satisfy frictional constrains can be obtained autonomously by adjusting the weighting factors of the normal forces of individual fingers. By incorporating continuous gradient flows, the optimal grasping forces of individual fingers for stable grasping can be quickly determined in real-time. Simulation results demonstrate the efficiency of the proposed method.

Key words: automatic control technology, robot, multi-fingered grasping, gradient flow, force optimization