In the past, motion tracking has often been a meticulous task. Complicated, time consuming systems with large arrays of sensors were needed to have any hope of capturing realistic human movement.
Now – according to a research team led by Professor Sungho Jo in collaboration with Professor Seunghwan Ko – complex motion capture systems may soon become a thing of the past.
Using a small sensor system placed on the wrist, the research team were able to accurately map the movement of the hand and fingers.
This confusing task was achieved by mapping the signals captured on the sensor using metal nanoparticle film on the wrist. Cracks in the film are picked up using laser technology. Then, using a form of deep learning, the system trains itself in real time, utilising a 3D hand to mirror the movement it believes is being captured.
Apart from the small size of the sensor, another major benefit of this new system is the flexibility. According to the team, the sensor can placed in various positions and angles on the wrist without negatively impacting it’s ability to sense motion. This is likely possible due to the systems ability to retrain itself.
The project spawned from the idea that motion tracking could be made easier by using a single point, rather than relying on multiple sensors placed all over the target.
Professor Sungho Jo says that the system can be expanded to other body parts in the future, potentially allowing the motion of the whole body to be captured using a relatively small network of sensors.
Link to download the full-text paper: