The prototype, developed by researchers at Carnegie Mellon University’s Human-Computer Interaction group, uses tiny impulses to sense wrist muscle movements. Operating on a principle of Electrical Impedance Tomography, similar to PET and CT machines, eight metal contacts send a continuous series of tiny electrical impulses through the wearer’s wrist. Sensors measure the strength of the impulses on the other side and, over time, collect enough data to generate and analyze a digital image of the finger and hand gestures being performed.
The wristband may not be the first capable of recognizing hand and wrist gestures, but it could be the most practical. Camera-based modules are typically much bulkier, the researchers note, and accelerometer and gyroscope-based systems can’t support static gestures. Moreover, the internal nature of Tomo’s analysis means it should work through gloves and clothing. And it’s cheap ($40 in its current form), small, and has a power draw minimal enough to make “integration directly into a smartwatch” feasible.
Tomo’s not perfect, though. Differences in wrist size and thickness between wearers mean the strap’s recognition software must “learn” gestures before it can accurately recognize them. (Think training voice dictation software to recognize diction.) And thanks to the granularity of the muscle measurements, the strap’s accuracy is largely dependent on its tightness and the consistency of its positioning.
Shortcomings aside, the Tomo’s the best attempt yet at a single-handed way of controlling a smartwatch. For the sake of those who never have a hand to spare, here’s hoping it catches on.
