Collision avoidance provider Iris Automation has received $8 million funding to bring its AI technology to autonomous drones and other flying vehicles.
In October 2015, Iris Automation’s founders put fingers to keys and wrote the company’s first lines of code. Two months later, they had a prototype for a system that aims to tackle the three big challenges facing the industrial drone industry: reliability, safety and autonomy. Their collision avoidance system for commercial drones promises to unlock the potential of beyond line-of-sight use cases.
After humble beginnings, the San Francisco-based company was accepted into the Y-Combinator start-up accelerator program in 2016 and, just yesterday, announced $8 million of Series A funding, led by Bessemer Ventures, to bring their AI-powered technology to life.
“Iris’ exceptional team has unlocked a $100 billion global industry by ensuring aviation safety in a world serviced by drones. No other technology comes close to their system in providing situational awareness in a feasible package for the flying robotics industry,” said David Cowan, the partner at Bessemer who led rounds in aerospace innovators like Skybox, Rocket Lab and Spire Global.
The operation of UAVs beyond line-of-sight without suitable sense and avoid systems is largely forbidden by regulatory bodies. That’s why, if drones are to become more widely useful, autonomous tools for industry, they need intelligent situational awareness technology.
Read more: Can drones and commercial aircraft safely share airspace?
The case for autonomous drones
Regardless of how skilled a drone pilot may be, it can be difficult to see hazards, due to the reliance of onboard cameras at longer ranges. Collisions can have expensive, or even tragic, consequences. When a company can’t demonstrate adequate mitigation of operational risk, regulators must limit drone applications.
Iris’s solution combines computer vision and deep learning algorithms to allow the drone to see the world much like a human pilot does – identifying potential hazards and intuiting speed and distance. The plug-and-play system interweaves basic deterministic algorithms with more advanced, nondeterministic algorithms and neural networks, which allow the system to be fault tolerant.
Harnessing this technology, the platform can detect distant objects, identify it (as a light aircraft, for example) and estimate its distance. A sophisticated logic core then autonomously manoeuvres the drone (or other flying robot) out of a collision trajectory.
Read more: Drones offer a new perspective on British Columbia wildfires
Rising above the competition
“Iris Automation’s approach to sensing is unlike anything ever attempted in the autonomous vehicle space,” said Alexander Harmsen, CEO at Iris Automation. “Our team of experts in computer vision, machine learning, and traditional aviation have built a product that will provide the level of safety necessary for pushing the boundaries of what is possible with drones, at a size factor and price point unheard of in the world of aviation.”
For all the artificial intelligence and machine learning power behind of its software, it’s somewhat surprising to learn that the hardware employs smartphone camera technology – but given the economies of scale of such devices, the cost, weight and power-intensiveness of radar, and the low resolution, unproven potential of LiDAR, it’s a shrewd choice.
“With a range of over 1,500ft, our system is 50x more powerful than the ‘bumper solutions’ that some current drone companies are using today with a mere 30ft detection range. Furthermore, the product is a standalone unit, agnostic to all platforms and can be integrated into any commercial drone in the world,” revealed Alejandro Galindo, Head of R&D.
Read more: Parrot revamps consumer drones for commercial market
Autonomous drones: the business applications
Industrial drone operations using pilots would not be economically viable for many companies. The more cost-effective ability to fly autonomously and BeyondVisualLineofSight (BVLOS) would pave the way for new methods of pipeline inspection, package delivery, large agribusiness, mining exploration, and much more besides.
Currently in open beta, Iris Automation is looking for partners in the UAV and drone space for its Early Adopter Program. Regulatory exceptions for BVLOS flights in the US Federal Aviation Administration’s (FAA) UAV rules (referred to as Part 107) allow drone companies to trial new technologies. There’s an opportunity to apply for BVLOS operations through the waiver process by using the Iris System before it is commercially available.
With its latest funding, Iris Automation is looking to expand its team (which already boasts expertise from the likes of NASA and Boeing) and scale its technology to participate in the upcoming White House UAS Integration Pilot Projects. Given, their product, talent and increasing financial backing, we could soon see myriad Iris-equipped autonomous drones rising out of Silicon Valley.
Read more: DJI launches FlightHub for drone fleet management