A Bio-Inspired Artificial Tactile Sensing System Based on Optical Microfiber and Enhanced by Neural Network

Human tactile perception involves the activation of mechanoreceptors located within the skin in response to external stimuli, along with the organization and processing within the brain. However, human sensations may be subject to the issues related to some physiological factors (such as skin injury or neurasthenia), resulting in inability to quantify tactile information. To address this challenge, a novel bio-inspired artificial tactile (BAT) sensing system enabled by the integration of optical microfiber (OM) with full-connected neural network (FCNN) in this paper is demonstrated, inspired by human physiological characteristics and tactile mechanisms. In this system, the BAT sensor mimics human skin, where the OM serves as the mechanoreceptor for sensing tactile stimuli, while the FCNN functions as a simulated human brain to train and extract the signal characteristics for intelligent object recognition. The experimental results indicate that the proposed BAT sensor can sensitively respond to both the contact force (static tactile stimuli), as well as the vibrotactile events (dynamic tactile stimuli) for the recognition of regular textures. Furthermore, by integrating the trained FCNN, the BAT sensing system accurately identifies various intricate surface textures with an exceptional accuracy of 95.7%, highlighting its potential in next-generation human-machine interaction and advanced robotics.