Ultrasound-based gesture recognition

| |
1 Star2 Stars3 Stars4 Stars5 Stars

Speech is probably the most natural interface between man and machine. However, rather “unhuman” solutions are being offered for dealing with loud ambient noise.

On January 9, 2007, Apple CEO Steve Jobs presented the first smartphone to include a touch-sensitive screen, thus sounding the death knell for numerous keys, buttons, and switches. These days, however, this “manual” interface between man and machine seems to be increasingly replaced by speech.

Both operating concepts have at least one thing in common: they are not always practical. For example, experts consider touchscreens in vehicles to be more hazardous than buttons, since the lack of haptic feedback reduces a driver’s attention to traffic, for example. And voice control fails in a convertible when the soft top is open.

Researchers at the Fraunhofer IPMS are therefore working on safer, non-contact approaches for communicating with robots, as well as in operating areas, vehicles, and households. At the heart of these efforts is a microchip that generates and receives high-frequency sound waves up to 300 kHz. If these are reflected in between by a finger, for instance, a measurement of the elapsed time as well as the phase and frequency shift according to the principle of the Doppler effect can be used to determine the position of the finger and its direction of movement.

For gestures over distances of up to half a meter, the devices achieve spatial resolutions in the sub-centimeter range. Compared to camera-based methods, ultrasonic transducers also enable the construction of significantly cheaper electronic and software systems due to longer signal transit times. In addition, they are not susceptible to stray light and allow for reliable data acquisition on optically transparent surfaces as well. And, last but not least, the CMOS-compatible systems can be produced in large quantities at low cost.

MEMS for gesture recognition

For this development work, researchers are using a new class of electrostatic micro-electromechanical (MEMS) bending actuators, which have been constantly evolving for the generation of audible sound in micro speakers and for micropumps since 2016. The proprietary nano-e-drive (NED) principle from the Fraunhofer IPMS leverages the high forces of electrostatic fields in nanometer-sized electrode gaps to allow for mechanical movements with displacements in the micrometer range. The chip surface as well as the complete volume is used for sound generation. This allows tiny components to be produced, with hundreds of them fitting on a single wafer. This reduces production costs when large quantities are involved.

From the perspective of the researchers, possible fields of applications for ultrasound-based non-contact motion detection include uses in automation, safety, and medical technology as well as the automotive and entertainment and household electronics industries.





Ultrasound-based operating concepts (Image: Fraunhofer IPMS).

Ultrasound-based operating concepts also work in loud vehicle interiors. (Image: Fraunhofer IPMS).