Gesture Recognition System Based on Motion Capture Suit and LSTM Neural Network

Abstract

Human-robot interaction (HRI) is the study of how humans interact with robots and how to design robots that can interact effectively with humans in a wide range of environments. The aim of this work is to design and implement an HRI system that uses motion capture suit and artificial neural networks to recognize dynamic gestures for improved user interaction. The design of the system takes the form of a simple graphical user environment application, in which the user sees his own movement in real time, also with the prediction of the gesture performed. The work laid the theoretical and practical foundation for future research in the area of reading motion data from mocap suits for the purpose of dynamic gesture detection. The contribution can be used in the design and implementation of other HRI systems in the future.