The Future of BVLOS and Drone Automation: An Interview with Alexander Harmsen, CEO of Iris Automation
BY Zacc Dukowitz28 May 2019
Iris Automation is at the forefront of creating technology to enable Beyond Visual Line of Sight (BVLOS) for drones.
Their AI-based collision avoidance system takes a novel approach to the problem, with a solution that allows a drone to “see” the world and react to it, just as a pilot would. The technology detects other aircraft, uses machine learning to classify them, makes intelligent decisions, reacts promptly, and is built to understand the aviation environment around the drone.
The technology has also been thoroughly tested—they’ve flown over 7,000 test flights—and they’re partners on multiple UAS IPP sites as well as on NASA’s UTM project. And just recently they launched Casia, the first commercial offering of their BVLOS solution for drones.
Alexander Harmsen, CEO of Iris Automation
We wanted to sit down with Alexander Harmsen, the CEO of Iris Automation, to learn more about how Casia works to enable BVLOS and to hear his thoughts on the future of drone automation.
Describe what Iris Automation does in one short sentence.
Iris Automation is an artificial intelligence and safety avionics company building collision-avoidance systems for autonomous vehicles, such as industrial drones.
How did you first get involved in the drone industry?
In college, James, my cofounder, and I started a UAS team that competed in national drone competitions across Canada. Shortly thereafter, I was the first software engineer at the drone package delivery company Matternet—this was very early on in the days of commercial drones.
Both of these experiences showed me how accessible these tools were, and what sort of widespread impact they were likely to have in the world in the following few decades.
Photo credit: Iris Automation
Tell us about Casia, the AI-based collision avoidance solution that Iris Automation recently launched. Why is it important and how do you see it impacting the drone industry?
Casia is the first commercial offering of Iris’ core computer vision collision-avoidance technology.
It’s an ultra light hardware and software solution designed to give UAS-makers and operators an all-in-one, turnkey system for Beyond Visual Line of Sight (BVLOS) operations.
Why is BVLOS so important for the future of the drone industry, and how could wide-spread permissions for BVLOS change the industry?
Casia will truly revolutionize the world by enabling the utilization of UAS in a way that the industry has been dreaming about for years, finally creating real value in the commercial use of drones.
Right now the use of drones is inefficient, requiring human pilots to be in the loop, having “eyes on the aircraft.” Having to keep a drone within the visual line of sight limits flight range to a mile or two from the pilot, and ultimately reduces the real potential of drone technology, making it an ineffective solution for most use cases.
By enabling UAS to fly BVLOS safely, Casia is the critical missing piece that takes commercial drone operations from theoretical/limited to functional/widespread. With Casia, UAS can properly be used for search and rescue, wildfire, medical package delivery, pipeline monitoring, agricultural surveys, forest management, and hundreds of other real-world examples.
How does Casia’s collision-avoidance technology work?
The system enables drones to see the world and react to it, just as a pilot would.
Casia scans the environment once every 200ms, creating a real-time map and looking for any potential intruders, all onboard. Once it detects another aircraft, it tracks it, and uses machine learning to classify it, estimating the risk profile, sending an alert to the ground station, and triggering an automated maneuver to avoid the collision.
To create robustness, the core system is composed of both AI and geometric computer vision algorithms. For the first time in the industry, the system allows a UAS to truly understand the aviation environment around it as if a pilot were on board.
Photo credit: Iris Automation
How is Casia being used in the field right now? Does a trained drone pilot always monitor the AI-based collision avoidance technology while in flight, or have you tested scenarios where the pilot is less involved?
The Casia technology has been extensively tested, with 7,000+ real-world test flights and mid-air collision scenarios—flying various manned aircraft against UAS—and over 40,000 encounters in simulation.
Over the last 2 years, Iris Automation has also run a successful early adopter program with more than 30 participating beta customers from five countries. These tests are a combination of alerts and manual intervention, and automated avoidance actions, giving us confidence in either approach.
Do you envision a future where drone flights (BVLOS or otherwise) are all fully automated—as, say, Uber might be imagining right now for self-driving cars—or do you think we’ll always need drone pilots to be involved in some capacity?
I think there’s no reason to think that in the near future most drone flights won’t be fully automated.
Similar to self-driving cars, these vehicles will behave predictably, will be always on (i.e. no falling asleep at the wheel), and will follow a specific set of rules similar to right-of-way rules in the sky and road today.
Ultimately, this means that the skies and roads will be much safer. As the use cases expand, these drones will do harder and harder tasks, expanding the potential of the technology and the impact it can have on society.
That being said, I do think there will be a role for humans to play in interacting with these systems (i.e. building, managing, deploying, analyzing) for at least the next 50 years.
Currently FAA waivers for BVLOS operations require a rigorous application process, which includes an in-depth review of specs related to all hardware in the proposed mission(s). Is it a goal of yours to have solutions like Casia on a list of accepted technology that is pre-approved for BVLOS, so that the entire process could potentially be fast-tracked?
Yes, that would be the ideal.
For the last few years we’ve been working quite closely with standards bodies like ASTM and RTCA to be able to develop a set of rigorous standards that can apply to these systems.
Additionally, we have done extensive testing with aviation regulators like the FAA and Transport Canada to show system performance, environmental testing, and set up a process for approval of scaled BVLOS operations for our customers all over the world.
Iris Automation has some impressive partnerships, including involvement in several UAS IPPs and NASA’s UTM project. Why have these partnerships been prioritized?
You’re right, we have spent a considerable amount of time over the last few years working to prove the reliability of our technology as part of government flight programs.
It is quite an investment, and ultimately worth it if it means that this technology can reach widespread adoption. The FAA IPP program in particular has been useful as our involvement has spread across five states (Oklahoma, Alaska, Nevada, Florida, and Kansas), and has given the regulators insight into how the system performs in many different environments, in different conditions, with different types of aircraft.
Photo credit: Iris Automation
The focus at Iris Automation seems to be primarily on enabling BVLOS through collision avoidance automation. Is there anything about the work your company does that people may not understand or know about, but which you believe to be crucial?
I believe that while our technology is crucial for enabling BVLOS, it is not the only line of defense.
For a system like this to work, there are other pieces of the puzzle that add to the total risk mitigation strategy. Recently we started building out integrations with qualified resellers, and working together with ADS-B and UTM manufacturers to make sure that there is capability across the entire stack to enable safer and more efficient operations.
What are your predictions for the drone industry? Please feel free to answer at length (what you see way down the road, what you see for next year, where you see regulations headed in the U.S. and/or elsewhere, new applications, etc.).
I still think that we’re at the infancy of the drone industry.
Right now, insurance and photography are core use cases for commercial users. In five years, most of the industry will be focused on all the newly unlocked BVLOS use cases, including search and rescue, mining, wildfire prevention, package delivery, pipeline monitoring, agricultural surveys, forest management, and many other use cases we haven’t even considered yet.
Before long, the investment that’s currently going into air taxis will pay off, and we’ll see Iris technology helping to ferry cargo, and also people within and in between cities all over the world. Unlocking this third dimension will drive down transportation costs, increase safety in public infrastructure, increase productivity across many industrial sectors, and continue to make travel and movement in general more accessible for more of the population.