Design, Development and Calibration of a Lightweight, Compliant Six-Axis Optical Force/Torque Sensor

Abstract

This paper introduces the fabrication of a six degree-of-freedom force and torque sensor based on fiber-optic sensing technology and its novel calibration methodology. The sensor is cost effective, lightweight, and flexible with a large force and torque measurement range suitable for biomechanics and rehabilitation systems particularly when a wearable sensing system is desired. Six fiber-optic sensing elements are used to detect three main forces Fx , Fy , and Fz , and three main torques Tx , Ty , and Tz . Sensor data were collected by applying dynamic forces and torques with various magnitudes, directions, and frequencies and compared with measurements obtained from a standard force and torque reference. The proposed calibration procedure is intended to reduce errors stemmed from a nonlinear force-deformation relationship and to increase the estimation speed by splitting the calibration into two estimation models: a linear model, based on a standard least squares method ( LSM ) to estimate the linear portion, and a nonlinear decision trees’ model ( DT ) to estimate the residuals. Both the models work simultaneously as a single calibration system named least squares decision trees LSDT . Using LSDT , the estimation speed increased by 55.17%, and the root mean square errors ( RMSEs ) reduced to 0.53%. In comparison, each model separately had a RMSEs of 1.26% and 4.70% for the DT and the LSM , respectively.