Towards Seamless Integration: Adaptive EMG Signal Processing for Avatar Control in Metaverse
Abstract
The next era of networks represents a set of metaverses connected into a single networked universe—the multiverse. These global challenges are increasingly attracting the interest of scientists and engineers. The International Telecommunication Union (ITU-T) began actively implementing an international standard in 2025, based on the recently adopted Resolution 105, which encourages and implements standardization of all aspects of the multiverse (avatars, protocols, platforms, network issues, etc.). One aspect is the method of interaction between users and their virtual avatars. Thus, telepresence suits have become part of immersive services, providing a maximum immersive experience for the user. In this framework, a distributed wearable network of micro-electromyography (EMG) sensors is employed. These sensors register low-amplitude muscle currents (perturbations) generated by the user. A long-standing problem in EMG signal processing is the robust extraction of the informative component of the signal. Addressing this problem is essential for improving the accuracy of user movement recognition and for accelerating EMG-based control of the user's digital avatar. The main aim of this work is twofold: (i) to review existing research on EMG processing in the context of telepresence systems and (ii) to propose an adaptive algorithm for EMG signal processing in telepresence suits that mitigates the influence of motion artifacts.