Preliminary analysis of the exploitation of QVAR sensor for gesture recognition

Recently, electrical charge (Q) variation (VAR) sensors have proven valuable in recognizing physio-muscular signals. In the act of driving, the muscle contraction of the hands provides information whether the driver is actually impressing forces on the steering wheel grip. This paper proposes the preliminary results of using QVAR sensor by STMicroelectronics, for gesture recognition, by measuring and processing the signal collected through wrist-electrodes to detect hand opening and closing. Another contribution is that the results are inferenced implementing a Decision Tree (DT) algorithm directly on the sensor side, the LSM6DSV16X, without any additional hardware, exploiting the use of the Machine Learning Core (MLC) provided by STMicroelectronics. The implementation of on-sensor intelligence paves the way for efficient deployment in scenarios where energy efficiency and privacy preservation are priorities.