Characterization Method for Bending Sensor Applied for Smart Glove

Nowadays accurate measurements of hand movements of workers in sectors such as manufacturing, aerospace, and healthcare are requested for different purposes, such as health analysis or human-robot interactions. In Industry 4.0, analyzing how workers interact with tools, machinery, and processes by hand it is important to achieve optimal performance, quality control, safety, and ergonomics. Several techniques - motion capture and wearable sensors - are available with different levels of accuracy and applicability. In particular, smart gloves, and in specific bend sensors, represent a viable solution to measure comfort level, but commercial solutions lack accuracy and affordability. For this reason, new sensors applied to wearable devices and therefore suitable characterization methods are needed. In this paper, a novel platform that emulates the finger movements is proposed to evaluate the sensors used to measure the rotation of two or more finger joints. For example, one bend sensor is used to measure more than one finger joint. The platform integrates a dummy little and index finger of average dimensions. With respect to previous works, the proposed method was used to test commercial bend sensors bent in two points, corresponding to two finger joints. The experimental results confirmed the sensor characteristics, especially regarding the linearity (the maximum error is less than 2.5%) and the repeatability (the maximum error is less than 3.8%). Finally, a relationship between the resistance and the curvature was found when the sensor was bent in two points, obtaining the same characteristics in terms of linearity and repeatability.