A single-element multimodal tactile interface with geometric signal encoding for robust robotic material and slip intelligence

In robotic tactile sensing, dynamic slip perception requires detecting contact state variations, sliding actions and interfacial properties. However, conventional flexible tactile sensors heavily rely on high-density spatial arrays to extract sliding information, leading to inherent limitations in achieving in-sensor self-decoupling of slip parameters. Here, we present a bio-inspired, multimodal tactile sensor that integrates spatially-programmed stepped electrodes for concurrent, all-in-one detection of normal pressure, sliding actions, and surface textures. The device features a vertically stacked architecture combining piezoresistive and triboelectric nanogenerator (TENG) mechanisms. While the piezoresistive layer ensures stable quantification of static normal forces, the centrosymmetric stepped design of the triboelectric stepped electrodes serves a physical encoder during dynamic slip events that enables hardware-level self-decoupling of sliding direction and velocity on a single sensing element. This design achieves 100% accuracy in sliding-direction classification, < 7% relative error in speed estimation, and < 3 mm absolute error in displacement measurement. Furthermore, assisted by a lightweight convolutional neural network (CNN), the fusion of multimodal signal attains 100% recognition accuracy across a diverse set of complex surface textures. This work demonstrates the immense application potential of such sliding-aware tactile sensors in intelligent robotics.