AI-POWERED IoT BASED SMART GLOVES FOR SIGN LANGUAGE RECOGNITION

Abstract

Many different kinds of human-computer and human-robot interactions may benefit from the clever and practical answer that is the ability to capture and recognize hand and arm motions (HRI).  Human gestures are a powerful means of interpersonal communication, and they also provide a natural and efficient means of interaction between people and technology.  The novel approach to real-time capture and detection of static and dynamic human gestures is shown in this study. This system measures data through strain sensors embedded in elastic gloves with ten sensors and stretchable bands with three sensors that measure strain.  These sensors should be fitted to specific joints of the human arm and hand to measure exact motion characteristics in the shoulder joint as well as elbow and wrist joints and proximal and metacarpal fingers. This research presents a new method that uses a radial basis function neural network (RBFNN) to detect and capture both static and moving human gestures in real-time.  By comparing the real-world data to a library of reference patterns that have already been collected, dynamic temporal warping (DTW) helps find potential candidates for dynamic behaviors and makes gesture detection easier.  The technique works well with both stationary and moving motions and with several joints in the target region.  Experiments including the recording and identification of both static and moving human hand and arm movements corroborate the efficacy of the suggested methods.