Abstract
Tactile sensing plays a crucial role in texture recognition, but variations in scanning speed pose a significant challenge for accurate discrimination. Previous studies have demonstrated that scanning speed alters the frequency of texture-induced vibrations, necessitating methods for speed encoding. In this study, we propose a bio-inspired spiking tactile sensing system that integrates mechanoreceptor responses with coincidence detector neurons to encode both texture and velocity without relying on external speed sensors. Our method enables speed and texture recognition in both active and passive touch scenarios by leveraging spike timing information from mechanoreceptors. We evaluated the robustness of our approach by introducing Gaussian noise into the neural encoding process, demonstrating that the model maintains stable accuracy with minimal degradation across different noise levels. The proposed artificial tactile system achieves an impressive 93% accuracy in jointly classifying texture and speed. Compared to prior methods, our model provides a biologically plausible solution to real-world tactile sensing challenges. This research offers a robust framework for texture recognition in prosthetic devices, robotic hands, and autonomous systems operating in unstructured environments.
