Predicting Linear Elongation With Conductive Thread-Based Sensors

Abstract

Conductive thread-based sensors have the potential to offer ambulatory measurement of body movement and posture. While such sensors have been previously described in terms of their overall correlation with existing measurement modalities, the ability of thread-based sensors to predict linear elongation has yet to be investigated. We constructed multiple stitched sensors, 20 cm in length, using two silver-coated nylon threads and seven different types of stitches, including cover and overlock stitches. Stitched sensors underwent elongation and cyclic stress tests while sensor resistance was recorded. Optimal predictive functions were derived for each sensor such that the range-normalized error was minimized. Responses were found to be governed by the combination of stitch geometry and physical properties of the threads. The average error in predicting elongation was 14% of the usable sensor range, suggesting that the stitched sensors are perhaps better suited to intermittent detection of stretched versus unstretched states rather than continuous prediction of absolute elongation. Nonetheless, coverstitches were generally superior to overlock stitches in terms of minimizing range-normalized error. The path of the conductive looper thread and the presence of stabilizing needle threads were important determinants of the sensor’s prediction capability and repeatability. Stress tests revealed a systematic upward drift of the resistance response curve for all sensors over repeated trials, suggesting that these stitched sensors require adequate recovery time between measurements. Given the possibility of applying stitched sensors non-destructively to existing garments, further study of these sensors—particularly those comprising multiple independent stitches in novel stitch geometries—remains worthwhile.