Taming the Variability of Soft Sensors

This study explored the use of a soft sensor made of carbon polymer (CP) on a smart wristband, to profile the subtle deformation of the wrist during hand gestures. To address these 3 issues using the CP sensors – (1) hysteresis, (2) varying response time and (3) variation in sensor, material decay, wrist size and wristband drift, a fusion-based folding point algorithm, a data collection strategy, and two calibration methods are proposed and experimented. 83%, 80%, 69% and 63% accuracy can be seen testing with data collected from the same session, from another session, from another wristband and from another user.