Shape optimal design and force sensitivity evaluation of six-axis force sensors

Abstract

A strategy for optimizing the design of a new type of wrist force sensor is presented. The design variables are the geometric sizes of the elastic sensor body, with sliding and rotating boundary conditions connected with the rigid rim. The calibration matrix, the condition number, strain-gage sensitivity, and strain gages glued to the sensor body are used to evaluate the performance of the force sensor. The finite-element method is employed here for numerical analysis of the elastic sensor body. The optimal design problem is solved by a penalty method determining the design variables. The natural frequencies and the von Mises stress are included in the constraint conditions in order to retain the force sensor within the higher natural frequency range and avoid failure, respectively. Based on the design criteria, a novel decoupled wrist force sensor is devised which, compared to existing commonly used force sensors, provides increased force sensitivity with minimum stiffness, the lowest condition number, and an excellent decoupled calibration matrix. The simple 6×6 calibration matrix affords genuine savings in calculation time and ensures the possibility of real-time control for a robot arm.