Neurala, With More Cash, Advances A.I. for Drones, Self-Driving Cars
Artificial intelligence software firm Neurala cut its teeth developing technology for government entities, including working on advanced navigation software for NASA’s planetary exploration robots.
Now, as A.I. technologies move further into our daily lives, 11-year-old Neurala says it is starting to pick up momentum selling its technology to businesses that are working in drones, self-driving cars, smart security cameras, and toys and consumer electronics.
Neurala’s customers embed its deep learning and computer vision software in their devices to “make their products more intelligent, interactive, and useful,” says co-founder and CEO Max Versace.
“We want to make this very easy and very fast and affordable for our customers, so it can jumpstart a whole new cycle of intelligent machines,” Versace says.
Investors are bullish on Neurala’s prospects. The Boston-based company recently closed a $14 million Series A funding round. Xconomy reported the investment in late December, based on an SEC filing, but Neurala is sharing more details about the deal this week.
Pelion Venture Partners led the round. Earlier Neurala backers 360 Capital Partners and Draper Associates also contributed, along with Sherpa Capital, Motorola Solutions Venture Capital, SK Ventures, and Idinvest Partners. Idinvest participated in the round through its Electranova Capital II Fund and in partnership with Ecomobility Ventures.
Neurala has raised $16 million to date from investors.
Neurala will use the new funds to develop new technologies, expand its customer base, and hire at least 10 more people over the next 18 months, Versace says. The company currently has 16 employees, plus another nine or so contractors, says Roger Matus, vice president of products and markets.
Versace founded the company with chief operating officer Heather Ames and chief technology officer Anatoly Gorshechnikov in 2006, while they were pursuing their PhDs in cognitive and neural systems at Boston University, the company’s website says. The idea was to create deep learning software that emulates how the brain analyzes its surroundings and learns. The founders developed a new way to perform parallel calculations on graphics processing units (GPUs), enabling inexpensive graphics processors to run sophisticated artificial neural network software and manage multiple sensors embedded in the device.
A retina-like feature in the vision system in Neurala’s software moves around and quickly classifies nearby objects, and can be trained by the user to recognize new types of objects in a matter of seconds, the company says. The software also enables machines to navigate their surroundings without GPS and avoid collisions by making calculations that try to anticipate the actions of nearby people and things, Neurala claims.
One of the main selling points for Neurala is its software can make all these calculations right on the device, without needing to connect to the Internet or a far-away server. That’s crucial for drones and autonomous vehicles, “where reaction time is critical and where you can’t always get fast network access,” Versace says. (Other tech companies are building similar capabilities into their intelligent machines, too. For example, Google’s experimental self-driving vehicles (now spun off as Waymo) handle at least some of the processing of their surroundings on board, according to a 2014 article from The Atlantic.)
That local processing capability also means that devices using Neurala’s software can sidestep privacy concerns that come with sending images and other consumer data to a company’s servers. That will no doubt be a relief for, say, parents buying high-tech dolls and other toys for their children.
“The reason this is really important is there’s sort of a branch of A.I. that I personally call ‘creepy A.I.,’ which is the A.I. that takes information, sends it back to a central server, and they mine it for details,” Matus says. “We don’t do creepy A.I. All of our stuff is local. We don’t have to send images of your children back to a central server for [the software] to work.”
Neurala, an alum of the Techstars Boston startup accelerator, has seen demand for its software pick up in recent months. Versace and Matus say the catalysts were attending a drone trade show last fall and releasing a software development kit, which makes it easier for developers to embed Neurala software in their products.
Neurala’s customers include Motorola Solutions, drone makers Parrot and Teal Drones, and a major automaker that the company declined to disclose. The car manufacturer is using Neurala’s software to advance development of autonomous vehicles, specifically in identifying and locating nearby objects, Matus says. Meanwhile, he says Teal Drones is using Neurala’s software for a consumer app that enables the drone to lock on to a designated individual or object and track its movements.
Matus says Neurala is also working with companies in Europe incorporating its software in drones that perform maintenance inspections of things like cell phone towers, power lines, and rooftop solar panels. The users can train Neurala’s software to recognize and flag damage and other issues.
“The drones can fly up to hard-to-reach locations and look for things, whether it be corrosion, or selected parts … and alert inspectors on the ground about what it sees,” Matus explains. “It adds intelligence, it adds speed, and it certainly makes it safer.”
Versace says it’s an exciting time to be working on artificial intelligence technologies, which are starting to move out of research labs and become real-world products. But there is still a lot that the technology can’t do.
“We are right at the beginning of the big explosion of the trend,” Versace says of advancing A.I.
Neurala has its work cut out as it goes up against other deep learning software companies, as well as big tech companies like Google, Amazon, Facebook, and Uber, which are all pouring significant resources into A.I. technologies.
“It’s a very competitive space, and a very hard space,” Versace says. “It’s nothing that scares us though.”
The key for Neurala, Versace says, will be investing and innovating in the right technical areas, and making sure those bets pay off with customers.
“We want to stay out of the science experiment,” Versace says. “We need to keep ahead, not only in terms of technology, but also in terms of translating that technology into tangible products.”