摘要
Visuotactile sensors have emerged as promising tools for digitizing the sense of touch, but most existing platforms remain limited to specific modalities such as high-resolution geometry, slip detection, or thermal cues, requiring reconfiguration or trade-offs in functionality. In contrast, human skin simultaneously perceives multiple modalities within a single surface. This work introduces OmniSense v2, a compact human-skin inspired visuotactile sensor that integrates multimodal tactile imaging into a unified gelatin-based elastomer layer. The elastomer incorporates UV-fluorescent markers for deformation tracking, thermochromic pigments for categorical temperature sensing, and a Lambertian surface coating for high-contrast texture imaging. These are paired with a programmable RGB/UV LED illumination system and a global-shutter vision module. Dedicated algorithms emulate the mechanoreceptive pathways of Merkel cells, Ruffini endings, Meissner and Pacinian corpuscles, and Aδ and C fibers through normal force mapping, tangential force classification, vibration measurement, surface imaging, and stepwise temperature sensing. Experimental evaluations demonstrate reliable normal force distribution mapping, >96% accuracy in slip, shear, and torque classification, vibration detection between 1–6 Hz, categorical thermal sensing across 15–75 °C, and a surface imaging resolution of 0.177 mm that surpasses human tactile discrimination (~0.4 mm). Stability tests showed functionality for six months, reusable calibration, and variability within ±5-12% across modalities. Latency was low for most tasks, with tangential sensing slower and higher variance (50-100ms). Furthermore, a handheld, time-multiplexed demonstration further illustrates near-simultaneous readouts of multiple modalities. These results position OmniSense v2 as a versatile visuotactile platform capable of capturing biologically mimicked tactile information across multiple modalities without mechanical reconfiguration.