Exploring the Potential of Machine Learning in Hand Motion Recognition Through Forearm EMG and IMU Data

Abstract

Machine learning and pattern recognition is an ever-evolving domain where major technological breakthroughs have occurred in recent times. This study aims to research the implementability of an innovative framework for hand motion recognition using forearm EMG and IMU recordings performed on real-life subjects. The literature review focuses on the background of those signals and on the characteristics that must be considered to measure and analyse them. The current status of hand motion recognition and its use within the control industry are also covered. The full process is then detailed, from the presentation of the material to the recording of the hand motions and data obtention. The establishment of models for the classification processes is presented and thorough analysis is performed. The main findings of this research are the great accuracy levels which are obtained as well while using full forearm recordings as measurement limited to the sole wrist area. These findings make it possible to imagine and suggest a future implementation of the technology in the shape of an innovation which could have a significant impact on the way systems are controlled. This research allows the opening of a new scope of study within the motion recognition world and may support future investigations that could result in breakthroughs and have a considerable impact on the relationship that humans have with the always more complex systems that surround them.