Estimating wrist joint angle with limited skin deformation information

Abstract

Ascertaining a person's motion intentions through muscle activity is important for controlling various assistive devices for people with disabilities. Several techniques have been proposed for estimating the extent of intended joint angle motion using skin deformation information derived from muscle contractions. The objective of this study is to verify our signal processing procedure for estimating intended wrist joint angle with skin deformation information in able-bodied subjects and subjects with an upper-limb amputation. Skin deformation was measured with a tactile sensor consisting of 48 distance sensors over a large measurement area. The root-mean-square error (RMSE) of the measured and estimated angles are evaluated offline using multiple linear regression in one individual with an upper-limb amputation and five able-bodied participants. In all tests, subjects undertook a wrist flexion and extension task guided by visual feedback, measured in real time. Sensors are selected in descending order of the standard deviation of each sensor's value. Strong relationships occur between the position and displacement of the area of greatest skin deformation and the intended wrist joint angle in all subjects. The minimum RMSE was 8.19° for the individual with an upper-limb amputation using 48 sensors as input, and 2.24° for able-bodied individuals using 16 sensors. One-way repeated-measures analysis of variance showed that at least 16 sensors are needed to reliably record skin deformation. Skin deformation analyzed with multiple linear regression is a plausible means of estimating intended wrist joint angle in persons with an upperlimb amputation. Even when a limited number of sensors (≥16) are used, continuous joint angle can be estimated reliably. These findings will inform the design of assistive devices that must noninvasively determine muscle activity.

An Investigation on the Sampling Frequency of the Upper-Limb Force Myographic Signals

“…For instance, hand squeezing activates the same muscle group as the power grasp, which is one of the most important actions we use in daily activities [45,47]. Palm pressing, which uses the same muscle group for wrist flexion-extension, is also a fundamental action used in daily activities and a key action included in many FMG investigations [9,20,23,48,49]. FMG was shown to be able to estimate finger movement status [22,33]; therefore, index finger pressing was included in the protocol to represent this fine finger action.…”

A Review of Force Myography Research and Development

“…It is also important to distinguish between the FMG approach and an approach that detects the change in forearm shape for gesture prediction. FMG uses sensors that detect the change in normal force due the change in stiffness of the MC, while the other approaches use camera [15], stretch sensors [16], strain gauges [17], capacitive sensors [18], or other types of transducers to detect the cross-sectional or overall displacement of the muscles during different limb actions [19]. The change in arm shape is also a result of the change in different muscles during contraction, but is not the same as changes in muscle stiffness.…”