Sistema de controle robótico baseado em leitura de sinais eletromiográficos

Abstract

This work presents the development of a control system for a robotic manipulator through surface electromyography (EMG) signals. The overall objective is to capture the electrical muscle activity of the human arm and translate it into specific robotic movement commands. The research is applied, experimental, and quantitative in nature, structured around the integration of instrumentation hardware, embedded systems, and digital processing algorithms. The theoretical framework is based on the physiology of muscle contraction, where the EMG signal represents the spatial and temporal summation of action potentials from physical movements. The methodology comprises non-invasive capture via surface electrodes, followed by analog conditioning of the tension, which employs an instrumentation sensor (AD8832) with a high common-mode rejection ratio (CMRR) to mitigate noise. The filtering stage uses band-pass filters (frequencies from 10 to 500 Hz) to isolate the frequency range of the highest energy in the signal and eliminate residual movement signals. After digitization via an analog-to-digital converter (ADC) in a microcontroller, the signal is processed for feature extraction, utilizing metrics such as Root Mean Square (RMS) and Mean Absolute Value (MAV) to quantify the user’s force intent. Pattern classification associates these values with thresholds that trigger commands sent to the robotic arm via communication protocols. The results obtained demonstrate the development of a system capable of reading the EMG signal, processing it in real time, and, through threshold calculation, determining and sending the corresponding command parameters for the robot's operation..