Simultaneous Pressure Sensors Monitoring System for Hand Gestures Recognition

The tracking and prediction of gestures present a high interest in many applications such as Prosthesis control, robotic tele manipulation, and rehabilitation. The common challenge thereby is the acquisition of suitable signals related to muscles constructions and to identify the corresponding gestures. In this paper, an measurement band based on 8 FSR sensors is proposed for the monitoring of the forearm surface force distribution as a basis for detecting muscle contractions related to gesture. A measurement system realizing simultaneous data acquisition of all sensors has been developed based on Bit-banging over a SPI communication protocol in a Raspberry pi 3 B+ board and 8 external ADSs. To build a data basis, ten healthy male volunteers were asked to perform 11 gestures belonging to American Sign Language numbers (from 0 to 10). For a real time classification, an algorithm is developed based on the Extreme Learning Machine method. The results demonstrate the feasibility of monitoring 8 sensor values simultaneously every 6 ms. The classification accuracy reached 90.09% for all tests.