Creating an effective gesture recognition systemwill significantly simplify the management of many modern gadgets. Developers from Switzerland and Japan have designed a compact device that attaches to a person’s hand and allows you to analyze gestures by changes in the acoustic characteristics of ultrasonic waves that occur when the brush or fingers move. At the same time, recognition of touching the fingers of the other hand and fixing the force of squeezing fingers is also available.
Modern devices are equipped with touchmonitors that control gadgets. However, in some cases, such control is difficult due to miniature screens that require physical contact with the user's fingers, but are not able to demonstrate large enough interface elements. Such devices include, for example, smart watches having a small display with a small number of physical control buttons.
A team of engineers from the Swiss HigherTechnical School of Zurich and the University of Tsukuba proposed an original solution that allows you to track gestures using ultrasound. The invention is based on the property of changing the shape of the brush with certain gestures, which leads to a change in the resonant frequencies of the brush. By correlating different resonant frequencies with certain gestures, you can create a control system.
The developers introduced two types of devices: fully autonomous running on a portable battery and connected to a computer. Only power and data transmission systems differ in systems (in the mobile version, information is transmitted via Bluetooth).
Meanwhile, the tracking sensor system,the fixing frequency change in both prototypes is the same and consists of two compact piezoelectric elements (25 × 12 × 6 mm), mounted on the outside of the brush with double-sided tape. One of the sensors emits ultrasound pulses with a frequency of 20 to 40 kHz, and the second receives a signal and detects a change in the resonant frequency.
In the process of analyzing data received from the microphone,they are converted into a vector, and then, using the support vector method, it is compared with various gestures. Scientists were able to develop recognition of various gestures, such as the movement of the thumb on other fingers or touching the phalanges. In addition, experts were able to identify the touch of the fingers of the other hand on the palm of the hand and determine the strength of the fingers.
Engineers reported achieving accuracyrecognition of gestures up to 84.4%, and the power of finger squeezing was identified with an accuracy of 85%. Currently, this technique requires an individual approach and training of the system for each person.
The invention can be used in systemsmanaging smart things, as well as when working with a virtual reality system. Engineers are now working on creating portable systems that are user-friendly.