Abstract
Simultaneous mapping of mechanical and thermal stimuli with flexible sensor arrays is complicated by signal coupling and pixel-to-pixel variation. Here, we report a scalable (300 × 300 mm2) photonic crystal hydrogel (PCH) sensor array integrated into a smart cushion, combining optical and electrical readouts with dynamic machine-learning signal decoupling. The sensor array incorporates a self-healing double-network hydrogel and shows a low interpixel color difference (ΔE < 5.6). It demonstrates high sensitivity for both mechanochromic (4.86 nm·kPa-1) and thermochromic (10.80 nm·°C-1) responses, respectively, and the sensor responses remained stable over 1000 mechanical loading cycles. Due to the different sensitivities of optical diffraction and internal ion migration, the visual signal resolves single-channel crosstalk by generating an independent data matrix that enables a dynamic support vector regression (SVR) algorithm to precisely decouple pressure and temperature variations (R2 = 0.996 and 0.980, respectively). Furthermore, by integrating a support vector machine (SVM) model, the system accurately recognizes distinct sitting postures (e.g., upright, forward-leaning, and asymmetric postures) in real time from the 25-channel spatiotemporal signals. The integrated PCH sensor array combines visual and electrical signals output to enable sitting-posture classification and thermal mapping, demonstrating great potential in human–machine interfaces and health monitoring.

This publication has 19 references indexed in Scilit: