Researchers at Cornell University have unveiled EchoMotion, a system that turns an ordinary smartwatch into a full-body motion tracker. The technology uses only the speaker and microphone already in the watch, combined with a lightweight artificial intelligence model, and needs no cameras, extra sensors or custom hardware.
It works like sonar: the speaker emits inaudible ultrasound waves that form a kind of sound bubble around the person, and the microphone picks up the echoes reflected by the arms, legs and torso. A deep learning model translates those echo patterns into the position of 12 joints, from head to ankles.
Test results
In a first test, 10 volunteers wore a prototype and performed 18 types of movement. The team then installed the system on an unmodified Samsung Galaxy Watch 6 and tested it with 15 new volunteers: without calibration, the average error per joint was under 10 centimetres. A single 12-minute calibration cut the error to about 6.1 centimetres.
The same model, with no retraining, also worked on a Galaxy Watch 7 and a Google Pixel Watch 3, with only a slight drop in accuracy. An exercise-tracking demo recognised 20 gym movements, such as squats and chest presses, with 95.4% accuracy — including in real gyms, with music at around 76 dB.
Efficiency and privacy
The system is light: the whole package is under 30 MB and ultrasound sensing draws about 12 milliwatts, roughly 40 times less than a wearable camera system. A 30-minute session on the Galaxy Watch 6 used only 5% to 6% of the battery.
Because it uses no camera, EchoMotion also protects the privacy of the wearer and of people nearby, and it generates less data to process. The main limitation observed was in busy hallways, where echoes from other people nearby reduced accuracy.
What comes next
For now, EchoMotion is a research project, not a feature available on watches for sale. The team is studying its use in workout tracking — counting reps and correcting form — and in physical rehabilitation at home. The researchers are also exploring combining the technology with first-person vision on smart glasses to track objects handled by the user.
The paper was presented on 13 October at the UbiComp/IWSC conference in Shanghai and published in the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies. The work comes from Cornell's SciFi Lab, directed by Cheng Zhang, with doctoral student Saif Mahmud as lead author. The research was supported by the US National Science Foundation.
Image credit: Samsung (press image) — Galaxy Watch 6, the watch used in the EchoMotion tests.
Sources: Cornell Chronicle; Digital Trends; Nanowerk.
Transparency: this article was produced with the assistance of artificial intelligence, cross-checking public posts on X and sources on the internet. Please consult the original sources to confirm the details.
Muse
