Flexible Dual‐Mode Tactile Sensor for Fine‐Grained Food Identification via Deep Learning

Equipping robots with multimodal tactile capabilities enables them to perform complex tasks; however, the development of flexible multimodal tactile sensors is often hindered by fabrication challenges and interference between integrated subcomponents. In this work, a dual‐mode tactile sensor for food identification is proposed, featuring an ultrathin structure composed solely of a flexible substrate and planar interdigitated electrodes. By combining impedance and triboelectric sensing under dynamic driving, the sensor captures signals that convey mechanical, surface, and compositional characteristics of food, enabling accurate identification. Building on this sensor, an intelligent food identification system is developed that fuses dual‐mode tactile data with deep learning algorithms. The system achieves a high classification accuracy of 99.26% across 15 common food items. It also demonstrates robust performance in distinguishing 10 types of carrots, covering three varieties and four freshness levels, with an accuracy of 94.62%. In addition, using impedance‐mode data alone, the system accurately identifies 12 types of vinasses with varying ingredient ratios, achieving an accuracy of 96.03%. These results highlight the sensor's effectiveness in both general food classification and fine‐grained quality differentiation tasks.