Low-Cost Embedded System Proposal for EMG Signals Recognition and Classification Using ARM Microcontroller and a High-Accuracy EMG Acquisition System

An electromyography signal (EMG) is known as a record of the action potentials that are produced by the human brain with the intention of activating or deactivating the skeletal area muscles. These signals are considered very important within the biomedical area because they contain information on muscle contraction and relaxation, which can be used in various areas. The EMG signals recognition and classification are a subject very studied within the scientific community for its various applications such as prosthesis management, virtual environment control, bio-robotics, etc. An EMG signals classification system fulfills four stages corresponding to the acquisition, preprocessing, processing and classification. In this system, last three stages, algorithms are used that in principle allow minimizing the interference produced in the acquisition of this type of signals, in second, they correctly extract the information or characteristics of the same signals, and in third, they allow to use techniques that allow classification in the best way the EMG signals. As technology advances, new needs emerge, day by day EMG recognition systems are in need of improvement, so new techniques are proposed that improve the processing and classification of these types of signals, but from the same In this way, these techniques require a higher computing cost that allows each one of the stages mentioned above to be fulfilled in the best way possible. Therefore, this article proposes a low-cost embedded system that allows EMG signals to be recognized and classified, as a signal acquisition technique, it is proposed to use a multi-channel high precision system acquisition, and for the preprocessing, processing and classification stages, it is proposed to use a high-performance, low-cost microcontroller such as the ARM cortex M4. The proposal in general aims to create a system of recognition and classification of EMG signal that can be used in different areas of biomedicine such as the design of prostheses, but with the particularity of an embedded system of high computational range and low cost, this will allow using the algorithms that emerge day by day to improve the recognition and classification of EMG signals.