How Self-Piloted Drones Will Avoid Collisions in the Air
A Vancouver startup is building a collision avoidance system for industrial drones.
The Iris Automation team with a drone. Image: Iris Automation
In June, two fighter jets had to be dispatched in a hurry when a suspicious-looking drone was spotted near the Ottawa airport. About two weeks later, a Quebec woman was hit on the head by a drone falling out of the sky, and was temporarily knocked unconscious.
With drones becoming increasingly popular, incidents like this are on the rise. Transport Canada's investigations of potential infractions involving drones (which could include anything from flying a drone where it isn't allowed to go, to flying recklessly and causing property damage or personal injury) increased from 61 in 2014, to 97 just one year later, a spokesperson told Motherboard via email. In fact, in the first five months of this year, the government agency has launched 28 investigations on potential drone infractions.
Drones are an impressive piece of technology, but they're essentially dumb machines: they typically need someone on the ground to pilot them, which limits how far (and where) they can safely go. While drones have become popular with hobbyists, this line-of-sight limitation has prevented drones from going very far on their own for industrial work, because of the concern that these high-flying robots may lead to dangerous accidents.
Vancouver startup Iris Automation proposes giving drones "eyes," through a fully autonomous collision avoidance system they've designed for industrial models. Its planned system has a detection range of over 500 m, significantly farther than what others can do.
The goal, they say, is safe flight beyond the drone pilot's line of sight.
The drone economy is booming. By 2020, drones are expected to evolve into a $100 billion market, according to one Goldman Sachs research analyst.
Iris Automation's goal is to create a fully autonomous, self-piloted drone that could help perform jobs like mining exploration, pipeline construction, agricultural surveys, package deliveries, forest management, search and rescue, and other activities.
It's already possible to get a drone from point A to B through a pre-programmed flight path, but drones are often incapable of autonomously avoiding collisions with obstacles that unexpectedly appear. Iris Automation hopes to solve this by creating a system that would give them "situational awareness," and fly themselves over long distances.
"These are huge, billion-dollar markets that are traditionally pretty slow to evolve. They have a lot of heavy machinery which could be replaced with drones," said Alexander Harmsen, CEO and co-founder of Iris Automation.
Their system, which is still under development, would allow drones to "see the world," he told me, and avoid collisions with trees, buildings, and mountains—but also with moving objects such as airplanes, birds, helicopters, balloons, crop-dusters and anything else that's flying around.
"We could have a fleet of 100 drones flying seven-to-eight hours a day over huge swaths of land"
"Think about mining surveys. Right now, you have helicopter pilots flying at crazy low heights, say 40 or 50 meters above the ground, for 10 hours a day. It's boring and it's so dangerous. We could have a fleet of 100 drones flying seven-to-eight hours a day over huge swaths of land, and not only is that cost effective, it is so much safer for the pilots," said Harmsen.
Harmsen started the company with co-founder (and CTO) James Howard, in their last year of school. The two met while completing their undergraduate degrees at the University of British Columbia. There, they started a drone team and competed nationally. "We were a ragtag team of students putting together projects on a shoestring budget. And the more we learned, the more we kept asking ourselves, 'What is holding back these huge industrial markets from using drones?' There is so much money here."
After graduating, Harmsen briefly had a co-op position with NASA's Jet Propulsion Laboratory, an Howard worked as an intern at Spire Global, a satellite-powered data company.
After doing a lot of customer demos, they raised a round of funding and got into Y Combinator, a prominent Silicon Valley startup accelerator that has backed the likes of Airbnb, Dropbox and Reddit.
Iris Automation isn't the only startup investing in collision avoidance technology.
In 2015, Intel invested $60 million in Hong Kong-based drone maker Yuneec. With help from Intel RealSense technologies, a Yuneec camera drone can avoid obstacles without a pilot at the helm. DJI Phantom also recently added a collision avoidance system to its Phantom 4 drone that can detect obstacles up to 15 meters away. But these are for smaller, consumer drones.
The startup hopes its technology will become the industry standard. Maybe soon, a self-driving drone that avoids crashes will mean we won't have to worry about drones falling on our heads due to operator error.
Correction: Harmsen and Howard started Iris Automation in their last year of school at UBC, not immediately after graduating. Also, Harmsen's work with NASA was on a co-op basis. The story has been updated to reflect both of these facts.