Remember the Batman echo-location smartphone system? That's real now

I've seen some pretty clever uses for smartwatch technology, but this one is straight out of a Batman film. Researchers have developed a way to track your entire body's movements using nothing more than ultrasonic sound waves and a smartwatch, effectively giving wearables their own form of echolocation.

Called EchoMotion, the system uses artificial intelligence to interpret sound waves bouncing off your body, working out where your joints are and how you're moving. Developed by researchers at Cornell University, it could eventually make full-body motion tracking possible without cameras, specialist sensors or complicated equipment.

Use what you've got

The clever bit is that it doesn't require a completely new piece of hardware, either. The researchers have managed to get EchoMotion working using the built-in speaker and microphone found in an ordinary commercial smartwatch, which could make it much easier to introduce into existing wearable technology.

The system works by sending out ultrasonic signals, which are essentially sound waves at frequencies too high for humans to hear. These bounce off the wearer's body before being picked up by the watch's microphone. An AI model then analyses the returning echoes to estimate the position of 12 different joints, covering everything from the head and shoulders to the legs and ankles.

It's a little like how bats navigate using echolocation, only instead of finding their way around a cave, the technology is building a picture of how someone moves.

Initially, the Cornell team developed its own 3D-printed smartwatch prototype, featuring additional speakers and microphones to improve the sensing process. However, the more impressive achievement came when it managed to transfer the technology to an unmodified Samsung Galaxy Watch 6.

In tests involving 15 people who hadn't previously used the system, EchoMotion predicted joint positions with an average error of less than 10cm. A one-off calibration session lasting just 12 minutes reduced that error to around 6.1cm. Even better, the same AI model worked on the Samsung Galaxy Watch 7 and Google Pixel Watch 3 without needing to be retrained.

Practical uses

There are some practical advantages beyond simply avoiding cameras, too. Because EchoMotion relies on sound rather than video, it doesn't need to record images of people exercising, which could make it useful in shared spaces like gyms. It also generates less data to process and is relatively energy efficient. During a 30-minute test on the Galaxy Watch 6, the system used just 5% to 6% of the battery, with its sensing process drawing around 12 milliwatts.

The potential applications are interesting. The researchers are already exploring how EchoMotion could be used to count exercise repetitions and assess technique, potentially allowing an AI fitness coach to spot when you're getting your form wrong during something like a Bulgarian split squat. It could also help monitor movement during physical rehabilitation, giving patients and healthcare professionals a better way to measure progress.

There are still limitations to overcome, however. Tests in different environments found that furniture and background noise weren't major problems, but the system struggled slightly in narrow spaces where echoes from other people interfered with its readings. The researchers believe that training the AI on data from thousands of users could help improve its performance in busier environments.

For now, EchoMotion is a research project rather than a feature you can activate on your existing smartwatch. Demonstrating it on several commercially available devices is a promising step towards making full-body tracking much more accessible. And if it eventually gives your watch the ability to understand how you're moving without needing a camera, that's a neat trick – even by Batman's standards.