Deep Learning-Enabled Auxetic Textile Sensors for Physiological Monitoring and Soft Robotics
Flexible wearable sensors have transformed motion tracking, soft robotics, and human-machine interfaces by enabling precise movement detection and adaptability to curved surfaces. However, conventional composite sensors often face challenges such as limited sensitivity, detection range, linearity, and durability. In this study, we propose a stretchable auxetic sensing textile with a negative Poisson’s ratio (NPR) structure, incorporating reduced graphene oxide (rGO) and carbon nanotubes (CNT) by micro-crack engineering to enhance its mechanical durability and sensing performance. Integrating macro-scale NPR with micro-scale wrinkles, this innovative design achieves a high sensitivity of 11.2 within a wide detection range (0-100%), a more linear sensing range with an R2 value of 0.998, an ultra-low detection limit of 0.5%, and exceptional durability, outperforming conventional wearable sensors. Additionally, the textile sensor boasts excellent moisture permeability (32.7 g m⁻² h⁻¹) and a remarkable NPR value of -0.25, ensuring comfort and adaptability for various wearable applications. Integrated with deep learning algorithms, the auxetic sensing textile demonstrates 98% accuracy in recognizing soft robotic movements at various bending angles. It is capable of capturing both small-scale physiological signals, such as electrocardiograms, and large-scale movements, offering significant freedom of movement and adaptability to complex surfaces.