A stretchable gold nanowire sensor and its characterization using machine learning for motion tracking

Motion analysis techniques are well-known for quantitative and objective human skills evaluation in various applications, including sport and medical skills training and development procedures. Such analysis invariably requires reliable data acquisition and analytical tools that can effectively monitor the human body movements. This paper introduces a novel soft and stretchable strain sensor for joint angle measurement applications. The strain sensor is made of ultrathin gold nanowires, which is highly stretchable, and sensitive. However, the electrical response of the sensor involves nonlinearity and hysteresis. To tackle this problem, we propose to use a machine learning algorithm to characterise the sensor for enhancing the predictability of the sensing response. A sensor embedded hinge system is developed for automated measurement of bending angle to validate the characteristics of the developed sensing system. The sensor has a high gauge factor of about 12 and shows high durability during a repeated stress-and-release test. It could measure the bending motion with an error of less than 2°. The developed sensor, combined with the proposed machine learning algorithm, can accurately monitor the bending motion. We believe that our sensing technology has great potential to be used for the evaluation and improvement of human skills proficiency.