Gravity compensation and output data decoupling of a novel six-dimensional force sensor

Abstract

A shunt three-legged parallel six-dimensional force sensor has been designed for more precise measurement of six-dimensional force/moment information. The theoretical static force model of the sensor was established based on the equivalent of a six-bar closed-loop parallel mechanism. The sensor has been experimentally calibrated under a given external load, and the neural network method has been utilized to nonlinearly fit the experimental data and achieve decoupling. Furthermore, a novel gravity compensation method for the six-dimensional force sensor of the wrist of a robot has been proposed based on the CAD variable geometry method. The positive solution of the position of the parallel robot is simulated through a wire-frame diagram, enabling accurate estimation and correction of the sensor. Experimental validation has confirmed the feasibility of the compensation algorithm.