Design of a Sign Language-to-Natural Language Translator Using Artificial Intelligence

Abstract

This paper describes the results obtained from the design and validation of translation gloves for Colombian sign language (LSC) to natural language. The MPU6050 sensors capture finger movements, and the TCA9548a card enables data multiplexing. Additionally, an Arduino Uno board preprocesses the data, and the Raspberry Pi interprets it using central tendency statistics, principal component analysis (PCA), and a neural network structure for pattern recognition. Finally, the sign is reproduced in audio format. The methodology developed below focuses on translating specific preselected words, achieving an average classification accuracy of 88.97%.