Co-founders Olga and Andrey Filippova are rethinking thermal imaging with open-source technology.
Elphel developed the first camera systems for Google Street View 15 years ago. The company has now set its sights on another game changer for night vision.
A PhD in physics, Andrey worked at the Prokhorov General Physics Institute of the Russian Academy of Sciences in Moscow before he and Olga emigrated to the U.S. in 1996. “Science was not in demand after the Soviet Union collapsed,” says Olga.
Andrey started working on a project with a collaborating company in Utah, planting the seed for their move. About five years later, the Filippovas started Elphel around open-source technologies. “We discovered the power of open hardware and free software, of Linux,” says Olga. “We started designing and manufacturing open-hardware cameras.”
An opportunity soon arose. “In 2003, Google approached us to manufacture and design the camera for the Google Books project,” says Olga. “It’s kind of an interesting story how we beat Kodak and Canon for this project.”
Canon’s mechanical shutter wasn’t up to the task, and Kodak had too long of a readout time. “Both companies were probably too large to jump ahead,” says Olga.
The Google Books project led to the camera systems for Street View, but Google ultimately brought manufacturing for both cameras in-house in 2008.
Elphel — short for “electronics/photoelectronics” — continues to supply camera systems to universities and labs worldwide, working with local contract manufacturer Newonics.
The business model shifted as camera technologies improved. “When smartphones developed and everyone had a camera in their pocket, that was really hard times for us, so we designed our own spherical panoramic camera like Google’s, but for private companies who couldn’t just use Google Street View,” says Olga.
“It lingered but didn’t grow big enough, and we went into 3D reconstruction based on images and developed our calibration methods. Still, even now, we are building our hardware, but mostly for scientific and research purposes. We design prototypes for our own research, and probably for the last four years, we’ve relied on government funding for research and development.”
Elphel won an SBIR grant from the U.S. Air Force to develop thermal sensors in 2019, followed by grants from NASA. The resulting technology, based on open-source hardware and software, has 20 times the contrast of the status quo, but the company now needs to develop the application-specific integrated circuit (ASIC) to move from prototype to finished product.
“We still use our core cameras,” says Olga. “We assemble the camera with thermal sensors instead of RBG sensors, or both, and we work on robust 3D reconstruction in the dark. They are pretty much standard sensors that FLIR or other companies manufacture, and with our system with multiple sensors, we increase contrast by about 20 times right now. Thas is a breakthrough in this technology, because thermal sensors cannot really, by the laws of physics, be much more effective than they are right now.”
Adds Andrey: “It’s applicable to autonomous cars, it can be for flying vehicles. . . . We’re trying to demonstrate the feasibility of this technology.”
Elphel is based out of a 3,000-square-foot facility in West Valley City. “We have a calibration room that is probably the largest in the world,” says Olga, noting that it occupies about a third of the total. “NASA probably doesn’t have calibration as advanced as our system.”
Challenges: “Hiring is a big challenge,” says Olga. “We need very, very smart people.”
There’s another looming barrier: Thermal vision hasn’t seen any innovation in decades, so potential buyers aren’t expecting disruption. “It’s been slow, incremental improvement,” says Andrey. “Nobody expected to get a breakthrough in this area, and it was very difficult to tell [people] that we have something that can really change this a lot.”
Opportunities: Elphel is focused on thermal sensing innovation. Andrey says the first big opportunity is developing a “killer application” for the defense sector, then moving into autonomous vehicles and aerospace.
A promising sign for the latter, Elphel was a winner of the 2022 Urban Air Mobility Innovation Challenge in September 2022. “For flying cars, you need to see it all in 3D,” he says. “LIDAR systems are naturally limited. Their range can be a little more, a little less, but it’s basically limited to about 200 meters, partially because of eye safety. If you use too high-powered lasers, they will be unsafe at short distances.”
Olga also points to “assisted driving” and augmented reality as potential markets for the company’s technology.
Needs: About $2 million to $5 million in funding to develop the ASIC. “To make a real commercial product, we need ASICs, so custom chips to run it, and development is expensive,” says Andrey. “We can’t make it ourselves.”
“We just need more exposure,” adds Olga. “We are very open to some kind of collaboration with others.”