Ultrasound is probably often called the tech that allows noninvasive physique scans and underwater communication, and will help us park our vehicles. A younger startup known as Sonair out of Norway desires to make use of it for one thing else: 3D laptop imaginative and prescient utilized in autonomous {hardware} purposes.
Sonair’s founder and CEO Knut Sandven believes the corporate’s utility of ultrasound know-how — a groundbreaking strategy that reads sound waves to detect individuals and objects in 3D, with minimal vitality and computational necessities — could be the premise of extra helpful and significantly inexpensive options than at the moment’s extra customary strategy utilizing lidar.
In info shared, Sonair has now raised $6.8 million in funding from early-stage specialists Skyfall and RunwayFBU, together with earlier traders, and is rolling out early entry to its tech. Initially, it will probably be geared toward autonomous cell robotic builders, however it’s imaginative and prescient (heh) is to see it being utilized in different purposes.
“Our go-to-market plan is to start out with robots, particularly autonomous cell robots [AMRs] — transferring stuff from A to B,” Sandven mentioned. “We’re beginning indoors — a limitation to provide us focus, [but] we are going to, after all, increase into the opposite robotic classes and into automotive in the long run.”
The title Sonair is a play on the 3D capabilities of water-traveling sonar, however utilized to sensors that learn alerts within the air — successfully what the startup has produced.
Sandven is an engineer and entrepreneur whose earlier startup, GasSecure, constructed fuel sensors primarily based on MEMS know-how — microelectromechanical programs combining mechanical and digital parts to create miniature gadgets with transferring components. (The petrochemical trade is a significant a part of Norway’s nationwide financial system.)
After GasSecure was acquired by a German industrial security specialist, Sandven began excited about different makes use of MEMS know-how, and regarded into analysis from SINTEF, a bunch that works with Norway’s most important science and technological college to commercialize analysis. Dozens of corporations have spun out of the group’s work through the years.
SINTEF had created a brand-new form of MEMS-related ultrasonic sensor, he mentioned, “which was prepared for commercialization.” Sandven acquired the IP for the know-how (particularly for acoustic ranging and detection), recruited the researchers who created it, and Sonair was born.
Though lidar has been a normal a part of autonomous programs improvement lately, there may be nonetheless area for complementary or various approaches. Lidar remains to be thought of costly, it has points with very, and nonetheless faces interference from mild sources, sure surfaces, and supplies.
Whereas corporations like Mobileye are trying extra critically at different alternate options like radar, Sandven believes Sonair has a viable probability, since its know-how will cut back the general price of sensor packages by between 50% and 80%.
“Our mission is to interchange lidar,” he mentioned.
The ultrasound sensors and associated software program that Sonair has constructed to “learn” the information from the sensors doesn’t work in a vacuum. Like lidar, it really works in live performance, with cameras to triangulate and supply extra correct photos to the autonomous system in query.
Sonair’s ultrasound tech is predicated on a “beamforming” methodology, which can also be what’s utilized in radar. The corporate says the information it picks up is then processed and mixed utilizing AI — particularly object recognition algorithms — to create spatial info from sound waves. The {hardware} utilizing the know-how (in preliminary instances, cell robots) will get a 180-degree subject of view with a spread of 5 meters and may use fewer sensors, whereas addressing a number of the shortcomings of lidar.
There are some fascinating concepts nonetheless to be explored right here. The corporate’s focus proper now’s on new strategies to enhance how effectively autonomous programs can perceive the objects in entrance of them. The tech, nonetheless, is small and in addition has the potential to be utilized in different kind of components. Might or not it’s, for instance, utilized in handsets or wearables as a complement or substitute for pressure-based haptic suggestions?
“What’s being performed at the moment by different corporations is [focused on] contact sensors,” Sandven mentioned. “After you contact one thing, the system will measure the strain or how arduous or delicate you’re touching. The place we are available is the second earlier than you contact. So if in case you have a hand approaching one thing, you’ll be able to reply already with our know-how. You’ll be able to transfer very quick in direction of the objects. However, then you definitely get precision distance measurements, so that you could be very delicate within the contact. It’s not what we’re doing proper now, but it surely’s one thing we may do.”
Sagar Chandna from RunwayFBU tasks that 2024 will see 200,000 autonomous cell robots produced, understanding to a market of $1.4 billion. That provides Sonair a right away market alternative as a less-expensive various for laptop imaginative and prescient parts.
“With diminished prices in sensor know-how and AI developments in notion and decision-making, industries from manufacturing to healthcare are poised to profit,” mentioned Skyfall accomplice Preben Songe-Møller.
