Invariant Neuromorphic Representations of Tactile Stimuli Improve Robustness of a Real‐Time Texture Classification System

Abstract

Humans possess an exquisite sense of touch, which robotic and prosthetic systems aim to replicate. Algorithms are developed to create neuron‐like (neuromorphic) spiking representations of texture that are invariant to the scanning speed and contact force applied in the sensing process. These spiking representations mimic the activity of mechanoreceptors in human skin and subsequent processing up to the brain. The algorithms are tested on a tactile texture dataset collected under 15 speed–force conditions. An offline texture classification system based on the invariant representations demonstrates higher classification accuracy, improved computational efficiency, and enhanced capability to identify textures explored in novel speed–force conditions. The speed‐invariance algorithm is adapted for a real‐time human‐operated texture classification system. In this system, invariant representations again improve classification accuracy, computational efficiency, and the ability to identify textures encountered under novel conditions. The invariant representation is particularly critical in this context, as human imprecision can be perceived as a novel condition by the classification system. These results show that invariant neuromorphic representations enable superior performance in neurorobotic sensing systems. Additionally, because the neuromorphic representations are grounded in biological processing, this work can serve as the basis for naturalistic sensory feedback for upper limb amputees.