Researchers at CTRL-labs, part of Meta’s Reality Labs, have developed a non-invasive neuromotor interface that uses machine learning and electromyography (EMG) hardware to translate wrist muscle signals into precise computer commands. Built with data from over 1,000 consenting participants performing various hand and finger gestures, this system enables intuitive digital device control without individual calibration, offering a practical solution for human-computer interaction.

The interface employs a wearable sEMG wristband to capture electrical signals from wrist muscle activity during movements like tapping, pinching, or waving. Machine learning models, trained on a diverse dataset, decode these signals in real-time to execute commands such as typing, clicking, or navigating interfaces. The system’s generalized model achieves high accuracy across users, with test participants demonstrating median performance of 0.66 target acquisitions per second in continuous navigation tasks, 0.88 gesture detections per second in discrete gesture tasks, and handwriting at 20.9 words per minute. Personalizing the models can improve handwriting performance by 16%, suggesting potential for user-specific adaptations.

Unlike traditional inputs like keyboards or touchscreens, the sEMG wristband is lightweight, portable, and unobtrusive, integrating easily into wearable devices like smartwatches. The system supports a wide range of tasks, from single-finger gestures to complex multi-finger combinations, making it versatile for various applications. It shows particular promise for assistive technologies, enabling individuals with motor impairments to interact with devices more naturally. Additionally, the interface could enhance gaming and virtual reality experiences by providing precise, gesture-based controls for greater immersion.

The researchers at CTRL-labs emphasize the system’s scalability, as it performs reliably across diverse user groups without requiring extensive setup. The high signal-to-noise ratio of sEMG allows for accurate detection of subtle gestures, overcoming limitations of vision-based systems, such as occlusion or lighting issues. This development highlights the potential of neuromotor interfaces to create seamless, inclusive, and efficient interactions, paving the way for broader integration into consumer electronics and specialized applications.