A research team at the University of California, Berkeley has developed a film-type device that allows AI to read hand gestures simply by wrapping it around the forearm. This film has a 64-point electrode, which AI analyzes the read neural signal pattern and compares it with previously learned information to determine what gestures are being made.
Until now, devices such as computer vision systems and sensor gloves that detect human hand gestures have been developed. However, in the case of computer vision, the arm in front of the camera had to be clearly visible, and the user’s face was reflected in the image, and sometimes it could not be said that there was no risk of privacy. The sensor glove requires a complex mechanism and has a problem that is fragile.
Considering this problem, the research team developed a film-type device that is wrapped around an arm. The user attaches the film between the elbow and the wrist and machine learns the various AIs to determine what it is like to make a gesture. It made it possible to identify 21 types of gestures, such as widening fists and palms and counting numbers by bending fingers.
In addition, AI always reviews and updates the gestures it learns. For example, it is possible to properly recognize and remember posture gestures that are slightly out of place due to sweat or posture gestures different from usual. In addition, data is processed on all devices, so there is no need for a cloud connection.
If this device technology is further developed, it can be expected to be put into practical use in applications such as electronic devices that manipulate with gestures, robot prosthetics capable of much more advanced movement than existing ones, and virtual reality controllers. Related information can be found here .