Inference System for Automatic Identification of Braille Blocks Using a Pressure Sensor Array
Inference System for Automatic Identification of Braille Blocks Using a Pressure Sensor Array
复制标题
使用压力传感器阵列自动识别盲文块的推理系统
DOI：
10.1109/percomworkshops53856.2022.9767257
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Tomonari Massuzaki
中科院分区：
文献类型：
--
作者：
Koichi Kuzume;Yoshitugu Watanabe;Haruko Masuda;Tomonari Massuzaki
视障人士在户外行走时，通常使用盲文字块作为引导标志。因此，准确检测盲文块对盲人的安全行走具有重要意义。在我们之前的研究中，我们提出了一种新的盲文块识别系统，主要是在户外环境中，不仅使用来自白色手杖的信息，还使用压力传感器阵列来更准确地实现盲文块。此外，我们还展示了我们的系统可以使用结构极其简单的神经网络来构建。本研究在前人研究成果的基础上，开发了一个盲文块自动识别的推理系统。该系统由16通道足底压力传感器阵列、微型计算机和云端系统组成，根据盲文块的类型分别检测压力模式的差异，对盲文块的类型进行推断，并实现机器学习。制作了一个原型系统，验证了该系统能够实时、高精度地推断出盲文字块的种类。
Visually impaired people usually use braille blocks as guidance marks when walking outside. Therefore, detecting the braille blocks accurately is very important for them to walk safely. In our previous research we proposed a novel system for recognizing braille blocks, mainly in an outdoor setting, using not only the information from a white cane but also a pressure sensor array to realize the braille block more accurately. Moreover, we showed our system could be constructed using a Neural Network with an extremely simple structure. In this research we developed an inference system for automatic identification of braille blocks based on the previous research results. The system consisted of a 16-channel foot pressure sensor array, a microcomputer and a cloud system, which respectively detected the difference in pressure pattern depending upon the type of braille block, made inferences about the kinds of braille blocks and implemented machine learning. We made a prototype system and confirmed that the system could infer the kinds of braille blocks with high accuracy in real-time.