Optimized Feature Extraction via Custom Low-Cost sEMG Hardware for Real-Time Myoelectric Interfaces

Surface Electromyography (sEMG) is a widely used modality to infer human motor intent in Human–Machine Interaction (HMI) systems. This study presents a cost-effective, custom-fabricated sEMG acquisition module with three differential channels strategically positioned to capture bioelectric activity from key forearm muscles. The analog front-end is designed with a high-gain amplification stage (2200×) and a bandpass filter spanning 48 to 482 Hz to suppress motion artifacts and ambient noise while preserving relevant myoelectric information. Acquired signals are processed through a lightweight digital pipeline including bandpass and 50 Hz notch filtering, followed by segmentation using a 200 ms sliding window with 50% overlap to balance temporal resolution and computational efficiency. Feature extraction focuses on commonly used time-domain descriptors (MAV, WL, RMS, VAR, and IEMG) and frequency-domain features (MNF and MDF). Experimental evaluation shows that the proposed system achieves a Signal-to-Noise Ratio (SNR) of 17.8–22.5 dB across three channels, while amplitude-based time-domain features (RMS and WL) exhibit less than 5% variability across repeated trials. These results indicate that the proposed hardware–software framework provides stable and reliable feature extraction suitable for real-time embedded applications. The system is therefore well-suited for wearable HMI scenarios such as myoelectric prostheses, robotic exoskeletons, and assistive robotic devices requiring low latency and low computational complexity.