Parameter Comparison of Grasping Force Optimization of Hard and Soft Multi-fingered Hands based on Safety Margin Detector

Abstract

In order to evaluate the influence of different types of finger contact models to multi-fingered grasping force optimization (GFO), two linear constrained gradient flow algorithms for hard finger (HF) and soft Unger (SF) hands are developed based on defined safety margin detectors. Firstly, based on the nonlinear friction cone constraints, two safety margin detectors for HF and SF are defined respectively, and their initial values can be obtained by iteration of force closure formula. It can be used as an effective initial point to start the gradient flow algorithm. Secondly, the gradient flow optimization algorithms with the safety margin detectors are developed for HF and SF types, which can real time detect the friction constraints and safety margin of each Unger at contact point, and provide early warning of the sliding trend in optimization. Take a three-fingered hand grasping as an example to compare and analyze the grasping force amplitude, safety margin and cost index under the same pose and external force wrench. Simulation experiments show that both two force optimization algorithms can eliminate the unreliable initial values before optimization, and optimize the force amplitude, safety margin and cost index convergence of HF and SF in continuous optimization. Under the same conditions, when the safety margin is positive at contact point, the SF hand can grasp with a smaller optimal grasping forces and obtain accurate force optimization results.