Interview: Ephraim Nowak, Co-Founder and Chief Engineer at Rain
When it Rains, it pours: Jon Adams spoke to Ephraim Nowak about innovative aerial firefighting solutions, utilizing UAS and AI for quicker and safer operations
How did you get involved in uncrewed aircraft systems (UAS)?
Rain evolved from a formative childhood experience. I grew up less than a mile from our CEO, Max Brodie, in the interior of British Columbia where, in 2003, the Okanagan Mountain Park firestorm caused the largest mass evacuations in Canada since the Second World War. Max often tells the story of helping his dad nail a soaker hose to their cedar roof, as plate-sized chunks of ash rained down and the police stood at the bottom of the driveway with a megaphone ordering them to leave.
In 2019 – in a backyard in Palo Alto – our Co-Founders Bryan Hatton, Max and myself reflected on this experience and started building the MK1 prototype, which became the world’s first fully autonomous demonstration of a drone system, detecting and suppressing a wildfire ignition in approximately 40 acres of forest.
Why did you feel drones would be an ideal firefighting solution?
A lot of people are working on early detection, mapping and prediction, but not many are going the step further to equip the aircraft to put the fire out. It seemed to us a natural follow-on, and a super-compelling one, once you realize how much safer and easier it is to put out a fire when it’s small.
The Rain Mk2 is an autonomous drone used to combat wildfires, based off an extant helicopter airframe. How is that beneficial when it comes to design and maintenance – and is that all done in-house?
We are developing Rain Aircraft Integration Kits that outfit existing aircraft with the tools that would be needed to fly autonomously, detect fires and extinguish them. Testing on the Mk2 was a proof of concept, that illustrated how getting a working prototype was possible.
Has the Rain Mk2 been created with other foundations in mind, so it can integrate with different technologies (software and hardware)?
As open architecture is being built, that allows us to collaborate with other agencies, whether that’s incorporating autonomous flight controllers or working with the ALERT Wildfire Camera Network. We’re not starting at zero with aircraft, either. Our goal is to quickly put out wildfires when they’re small, not to reinvent everything.
The Rain Mk2 has a 30-gallon payload, designed to rapidly contain ignitions before they get out of control; what retardant do you use and how do the aircraft refill their payload?
We’re currently using the same payload as many firefighting helicopters: a 99 per cent water solution with added foam concentrate. The deployment mechanism we developed for the Mk2 is the first of its kind – a Compressed Air Foam System gives up to a 30x expansion ratio. Coupled with our onboard perception and real-time fire-mapping software, this allows extremely precise deployment of foam, exactly where it is required.
The drones are a rapid response solution to allow the fire services time to reach the blaze and deliver better conditions to manage it. How are the drones informed of an ignition, and how long does it take before they are in the air?
Rain integrates with early detection methods such as the ALERT Wildfire Camera Network, utility Smart Grid data, lightning detection networks, and GOES satellites to obtain suspected ignition coordinates. Then, a Rain aircraft takes off from a nearby Rain Station and contains the fire within the first 10 minutes of that initial detection.
What role does AI play in their navigation, communication and ability to combat the fires?
As the aircraft approaches a suspected ignition, it compares predicted fire growth with actual fire growth and automatically dispatches other nearby Rain aircraft if required. Software on Rain aircraft uses thermal cameras and computer vision to precisely deploy fire retardant for direct or indirect suppression.
How many drones do you envisage being deployed to any one fire emergency? And will you keep them at a central hub, or be deployed from multiple sites simultaneously?
How many aircraft are deployed to any ignition is basically determined by our Installation Network Simulator, which takes into account the underlying factors that drive fire growth in any particular region. This informs the optimal placement of Rain Stations – the remote take-off and landing sites that house the aircraft. Rain Stations are optimally placed to guarantee coverage in any particular region and can house multiple aircraft. The decision to deploy one or multiple aircraft is made in real time, based on early detection information and prevailing conditions at the time.
How will the drones interact safely with conventionally crewed aircraft?
There is a crawl, walk, run approach to integrating into the national airspace. The wildland urban interface is the ideal location for deploying this technology. Our aircraft are designed to comply with ADS-B requirements and will initially be deployed with higher levels of human oversight. As the autonomy technology matures, we will be able to elevate the human operator to provide oversight, rather than being more hands-on.
Do you think the Rain Mk2 will lead to technology or designs that will be used for other services beyond firefighting?
Our focus is on containing wildfire – it’s a problem that everyone from agencies, local governments, utilities and even housing developers are really concerned about.
Your Mk2 is currently in the demonstrator stage; when do you think we can expect to see full-scale deployment and what’s next from Rain?
This year, we are working on a pilot project with the San Mateo County Fire Department. This demonstration will showcase a Rain response to a wildfire ignition in a large geographic area – in this case, a local county.