Design and characterization of a smart fabric sensor to recognize human intention for robotic applications

Abstract

Human intention recognition still represent an appealing challenge when seeking to design wearable exoskeleton for daily use. Even though biosignal-based sensors are the most widespread in this context, the innovation introduced by conductive smart fabrics can lead to overcome the biosignal analysis limitation related to motion artifacts and signal-to-noise ratio. This paper aims at designing and metrologically characterizing an innovative sensor based on smart fabrics to embed into upper limb wearable exoskeletons. The sensor was composed of two piezoresistive units, which are made with a smart conductive fabric, and a mechanical structure, which was 3D-printed with PC/ABS filament, to be integrated it in the exoskeleton. A test bed was realized to metrologically characterize the sensor in terms of linearity, repeatability, response time and hysteresis. The characterization was performed in both compression and traction conditions. A linearity error lower than 6% was found for both piezoresistive units, with an average sensitivity equal to 0.06 V/N and 0.08 V/N, respectively. A variability always lower of 5% was found for the measurement of 2 kg and 5 kg, assessing the possibility to use such sensor for the desired application. An average hysteresis error up to 20% was found, whereas time constant average values were found equal to 0.28 s and 0.73 s for the piezoresistive units sensitive to the compression and the traction, respectively. These findings confirm the possibility to use smart fabric for the recognition of human intention by exploiting the advantages associated with the cost, flexibility, easy data interpretation and non-invasiveness.