Smartwatch technology continues to shift away from simple activity tracking toward complex human-computer interaction. Researchers at Northwestern University have introduced EITWatch, a prototype system capable of recognizing hand gestures without the need for external cameras or bulky sensors. This device uses a method called electrical impedance tomography, which involves sending low-level electrical currents through the wrist to detect muscle and tendon movement beneath the skin.
The hardware implementation distinguishes EITWatch from previous iterations of this technology. While past systems required electrodes to fully wrap around the wrist, this prototype fits eight gold-plated electrodes within a 31 mm ring under a 40 mm watch case. The device uses a multi-depth scanning technique to reach deeper muscle groups, which improves the accuracy of gesture detection compared to standard surface-level measurement methods.
Testing involved 12 participants performing two distinct gesture sets. One set focused on six larger static hand poses, while the other included five subtle micro-gestures. The system demonstrated high performance in controlled tests, reaching over 91 percent accuracy for both categories. These results confirm that a watch-sized contact patch can collect sufficient data to identify specific manual inputs.
However, the technology faces challenges regarding long-term stability and cross-user consistency. Accuracy declines when users reposition the watch or when the system encounters data from new individuals without prior calibration. The research team is now focused on refining data representation methods to maintain recognition reliability across repeated use. Successful development could lead to a standard interface for smartwatches that does not require additional hardware or external accessories.