Integrated Sensing and Computing for Hand Gesture Recognition Based on Flexible Printed Electronics

This paper presents a tiny and flexible printed-electronics approach to hand gesture recognition based on piezoresistive sensing and integrated sensing and computing (ISAC). We fabricate printed piezoresistive sensing nodes for gesture data collection and design the corresponding analog computing network using a mature printed-electronics process design kit (pPDK). The main contribution is a pPDK-grounded ISAC architecture and a hardware-aware optimization algorithm for operating-point calibration, topology sparsification, and variation-aware training. Leave-one-subject-out experimental results across 20 subjects show an accuracy of 95.84% and a macro-F1 score of 94.63%. In the optimized circuit design, this performance corresponds to a projected requirement of 103 resistors and 32 transistors for the sensing-and-computing network.