Design and Implementation of a High-Precision Braille Recognition System Based on Biomimetic Multi-Convex-Points U-Shaped Micro-Nano Fiber

Abstract

As a vital medium for visually impaired individuals to access textual information, Braille recognition technology plays a crucial role in enhancing this group's ability to access information without barriers. To address the limitations of existing Braille recognition technologies based on optical imaging and piezoelectric sensing, such as bulky equipment, environmental sensitivity, poor flexibility, and susceptibility to signal interference, this study proposes and implements a highly robust Braille recognition solution based on a biomimetic multi-convex-points U-shaped micro-nano fiber (MNF) sensing system. We designed a bio-inspired multi-probe flexible sensing film inspired by the discrete pad array of animal footpads. This layer is integrated with highly sensitive U-shaped micro-nano fiber sensing units. This structure enables spatially separated acquisition of Braille dot matrix tactile signals, providing raw data with high resolution and low disturbance for subsequent algorithms, enhancing the quality of collected data. For the acquired high-quality raw time-series signals, a comprehensive embedded signal acquisition and processing system was constructed, and a recognition algorithm based on a stacked Long Short-Term Memory (LSTM) network was designed. Through systematic data preprocessing, network architecture optimization, and hyperparameter tuning, the algorithm can effectively identify the dynamic characteristics of sensors sliding over different Braille characters and provide predictive results. Experimental results demonstrate that the system achieves an overall recognition accuracy of 99.74% on an independent test set comprising 390 samples, demonstrating its high precision and reliability. This research not only provides an innovative hardware solution and efficient algorithm for portable, highly robust Braille recognition but also offers design insights and practical examples for the application of micro-nano fibers in dynamic tactile perception.