Soft and Highly Adhesive Wearable Electronics for Hand Reconstruction Based on PMUT and PPA‐CNTs Strain Sensors

Abstract

3D hand reconstruction is an advanced technology in human‐machine interaction (HMI), robotic control, and medical rehabilitation applications. However, methods based on optical cameras and data gloves suffer from high equipment costs, complex algorithms, susceptibility to ambient light interference, and high drift of inertial measurement units (IMUs). Here, a soft electronic skin (E‐skin) for 3D hand reconstruction, inspired by human biology is proposed, which integrates a multi‐sensor fusion of stretchable carbon nanotubes (CNTs) strain sensors and microelectromechanical system (MEMS) ultrasonic transducers. A straightforward screen‐printing process is introduced‐ to fabricate a multi‐layer stacked structure of the E‐skin. The substrates of both the E‐skin and strain sensors use the same material, a polyethylene glycol (PEG) mixed polydimethylsiloxane (PDMS) adhesive (PPA), which strengthens the bonding between the layers. The optimized PPA exhibits a low modulus (186 kPa), high elongation (>220%), and strong adhesion (1.2 N cm−1), while the PPA‐CNT composite strain sensor demonstrates excellent sensitivity linearity (0.99) and minimal resistance drift over 500 stretching cycles. The PPA material combines the waterproof and biocompatible properties of PDMS, while also achieving high adhesion and softness, allowing it to maintain conformal contact during finger bending (strain >70%) without glue or bandage. A Quantile Regression Neural Network (QRNN) algorithm is introduced to improve dynamic accuracy and robustness in finger joint angle detection. The system's application in gesture recognition and VR interaction is demonstrated, achieving high accuracy in sign language recognition and robust hand tracking.