Artificial intelligence (AI)-driven smart glove for object recognition application

Abstract

The development of soft sensors with high sensitivities and good response time is particularly useful for healthcare and robotics applications. This article presents a smart glove with an object recognition feature using the support vector machine (SVM), a supervised machine learning algorithm. A smart glove was fabricated using five resistive flex sensors, which can be attached on each of the fingers, respectively. The flex sensor was characterized. A microcontroller was used to receive and process the data from the flex sensor and transmit it to a PC for machine learning and prediction. The SVM model was trained using 160 training data set, and the trained model could be used to recognize three objects with different shapes with an accuracy of 91.88%. The proposed AI-based smart glove has shown high accuracy in object recognition. The proposed approach can possibly provide a promising low-cost solution for the healthcare and robotics industries.