Tactile Skin

High-resolution tactile skins usually get finer spatial resolution by adding more sensing elements, which increases wiring, readout, and structural complexity. This project presents a thin liquid-metal electronic skin that achieves super-resolution touch localisation and normal-force sensing from only four resistive channels, by encoding contact information directly into the geometry of interlocked liquid-metal microchannels.

A kernel-based fast simulator predicts topology-dependent resistance responses without repeated full-field finite-element simulations, enabling noise-aware optimisation of channel geometry for spatial distinguishability. A hierarchical mixture-of-experts (HMoE) decoder then jointly estimates contact position and force. Across three independently fabricated devices, the sensor achieved a mean localisation RMSE of 0.230 mm; a representative device reached 0.228 mm localisation RMSE and 0.043 N force RMSE, corresponding to a super-resolution factor of 58.3.

Force-modulated handwriting and a wearable pressure-tiered keyboard demonstrate continuous and discrete human–machine interaction, with the keyboard achieving 95.38% joint key-and-mode classification accuracy.