Launch HN: Radical (YC W23) – Autonomous high-altitude solar aircraft

402 points by jthomaslm ↗ HN
Hey, HN! We're James and Cyriel, co-founders of Radical. We're making an autonomous solar-powered aircraft designed for continuous flight in the stratosphere. Our 20 ft. prototype recently flew nonstop for over 24 hours, which you can see in this video: https://youtu.be/E6oDxQYEksc (edit—warning, there are rapidly flashing lights starting just after 1m, in case you're sensitive to that). Our website is at https://www.radicalaero.com/, but it's pretty bare. If you use LinkedIn, we have a few more posts at https://www.linkedin.com/company/radicalaero.

Our aircraft is designed to fly over specific areas for months, carrying various payloads for tasks like imagery, sensing, and telecommunications. What we’re building behaves like a drone in some ways, and like a satellite in other ways. Much like satellites, we’re able to provide service for long periods of time, but we’re also much closer to users (we fly at around 20 km altitude) and able to maneuver or remain over an area of interest. This makes what we’re building really well suited to applications that require continuous coverage or high-resolution/bandwidth data.

Examples of this include continuous real-time monitoring (such as in wildfire management or illegal fishing), high-resolution mapping and imagery (we’re able to collect sub-10 cm resolution imagery), and high-speed direct-to-device internet. The ability to permanently host sensors and devices in the sky in this way opens the door to lots of new opportunities. In truth, we still don’t know all the new applications that will arise from this (we’re really interested to hear your thoughts on potential applications!).

As for the technical details: Our aircraft is battery electric and driven by propellers. It has a large wing for high aerodynamic efficiency and to generate the necessary lift required to fly in the thin air of the stratosphere. The wing is covered in solar cells, during the day, these power the aircraft and charge its batteries. Through the night, battery energy is deployed to continue flight. We repeat this process daily, enabling us to fly for up to a year without needing to land. Ultimately, battery cycle life is what limits our aircraft’s flight endurance - and we can land, carry out simple maintenance tasks and then re-launch to continue flying. Our aircraft has multiple tails which help to stabilize the ultra-lightweight structure (our 20 ft. prototype weighs just 13 lbs.). We also use these tails to control the aircraft, they provide roll control by twisting the main wing of the aircraft, increasing or decreasing lift as needed.

The aircraft is fully autonomous; it has a full autopilot system onboard and various sensors for position, airspeed, and other key data streams (much like a typical drone or UAS). The aircraft flies at high altitudes of around 70,000 ft. (20 km) avoiding cloud cover, civil air traffic, and the turbulent winds of the troposphere.

Long endurance flight has been the goal of many past projects. NASA’s Helios and the DARPA Vulture program tried to develop long endurance aircraft. Helios’s crash led to an overhaul of aircraft structural analysis codes, and DARPA Vulture led to advancements in battery and solar tech. More recently, both Facebook’s Aquila and Google’s Loon were discontinued. Recent advancements in battery and solar tech, and miniaturization of electronics mean long endurance flight is now feasible - but we are aware we need to do things differently to succeed. Unlike those before us, we’re not tying ourselves to a single application and are focused on bringing a cost-effective solution to market. That means avoiding research-grade components and moving quickly. Additionally, we firmly believe that iterat...

205 comments

[ 3.2 ms ] story [ 238 ms ] thread
Love the demo video from the 24h non-stop flight :)

Congrats on the milestone & keep on pushing!!

Solar is awesome, so I like what you're doing.

Besides that "high-speed direct-to-device internet" caught my attention. Are you planning on competing with Starlink? Is 20km high enough to cover a big enough area?

There is definitely some market overlap in wireless telecommunications. But there are some distinct differences with satellite services. One big one is that our aircraft isn’t in orbit, so it’s much easier for us to provide targeted coverage over a specific area without needing a full constellation. Flying at 20 km also puts us much closer to users which makes high-speed direct-to-device service (5G) a lot easier.

In terms of coverage, from that altitude, we have good line of sight coverage over large areas. Obviously, satellites can cover wider areas, but it turns out that in most suburban/rural settings the ground footprint becomes limited by the population density and bandwidth available - so the larger satellite footprint is only really useful in extremely remote locations.

This is fantastic, congratulations! There are many use cases for a technology like this, even with a small and light payload (if I were at one of my previous jobs I would be sending you emails already about planning some tests). Do you envision being able to provide some power, however small, to a payload?
Definitely - in most cases we expect our payloads will draw some power. The power available depends whether you’re looking for continuous or periodic/peak power draws.
Was the 24 hour test at 20km? Doesn't it take a lot of battery power to climb to that height?

>we then spent 6 years working together on delivery drones at Amazon Prime Air

Was that ever a serious endeavour or just a publicity stunt?

The 24 hour test was with a sub-scale prototype at low altitudes. It turns out that testing a small aircraft in the thicker air at these altitudes matches some important physics of the larger aircraft (Reynolds number).

Perhaps unexpectedly, climbing to 20km altitude isn’t too large a problem from an energy perspective. In a typical energy cycle, the aircraft has low battery in the morning, so if we launch with a full battery charge we have plenty of extra energy to climb up to altitude.

Prime Air - definitely a serious project, but I’m sure Amazon didn’t hesitate in milking the PR! Making a reliable drone delivery service at Amazon scales certainly isn’t easy. That team is still going strong and I’m sure we’ll see more from them as they ramp up commercial deliveries.

>The 24 hour test was with a sub-scale prototype at low altitudes.

That's fine, but I think you need to make that clear or some people might think you aren't being straight.

Just over 20 years ago, a small model aircraft (https://en.wikipedia.org/wiki/The_Spirit_of_Butts%27_Farm) crossed the Atlantic in 39 hours using about a gallon of Naptha as fuel. The same year, the Design-Build-Fly club at my university built an electric aircraft which had some serious endurance and payload capabilities, though I don't remember the details. Keeping a model electric plane aloft for a day is impressive, but how many hobbyists have built something similar by now without any expectation of VC money?
Impressive. Did it rotate the propellor the whole 38.9 hours on 1 gallon of fuel?
This sounds phenomenal, always really glad to see folks tackling very ambitious challenges that solve serious problems! I think there are going to be many applications in the defense and security (port, border, maritime) domain for sure.
Nice. Who are you looking to hire?
We’ll be hiring in the near future. Early engineering hires will likely be controls/RF/EE focused, and we might bring on some biz dev support.
Any need of someone with a background in aeronautics/flight-simulation ;)
this looks amazing. Particularly hopeful/interested if this could be used for wildfire prevention. Great work!
Thanks, wildfires inspired us to work on this technology, so we're excited to see it used in those applications too!
Super cool!

I have purely cosmetic feedback about the website, which is that the tagline "from the stratosphere" doesn't match up with the images being played behind, which are pretty clearly low altitude. I am assuming that's because it was easier (or possible whatsoever) to film lower, but it gave me a little mental hiccup.

I look forward to seeing what y'all come up with!

Thanks so much, it's a lot of fun to work on.

Thanks for the website feedback - for the sake of clarity, our test flights with the sub-scale aircraft have all been at lower altitudes. The stratosphere is the next step and we plan to fly there next summer!

Wow, 10cm live view for complete situational awareness of forest fires feels like it should be a game changer.

Can we make sure someone (if not you guys) fully builds this service on your flights for the fire service?

The standard story arc for something like this is that it is pitched as a humanitarian/environmental technology, but it is the military that provide the money and that where it gets used. I'm not being overly judgemental - the good guys (e.g. Ukraine) need good military tech. But lets not kid ourselves.
Basically any technology that makes any difference will have a military value, and will be used by the military.

One of the few things that are not obviously, if in any way at all, usable for any military purposes is targeted advertising. Whether it's used by the good guys more is an interesting topic.

That's not actually true about ad targeting. The tech used there is a big information warfare thing, which has major military applications
I receive targeted recruitment ads from the U.S. Army and Navy frequently. Targeted advertising appears to be a valuable tool in encouraging young people to join the military.
Bad actors are using ad tech to subvert democracy (cf Cambridge Analytica).
Wildland forest fires were actually the inspiration for this technology! A year ago Seattle had the worst air quality in the world for a few days, which piqued my interest in the area. Speaking to people in the USFS we learned about the challenges of getting real-time high-resolution data, and realized that an ultra long endurance drone would be a great way to get this.
Any idea how big an effect air quality has on cell efficiency? Do cells work with diffuse light or is 20km high enough to be unaffected?
Flying at 20 km should place us above cloud cover and fire smoke, but the cells do work surprisingly well in diffuse light. Ensuring that we use imaging sensors that can cut through the smoke will be key (most likely LWIR or MWIR)
You know what would be really helpful in such a use case? Thermal imaging. Now I'll admit I know nothing about Thermal Imaging, but I imagine if it was possible to get a live high resolution thermal image of an area, you could identify hotspots quickly. Heck, even for SAR missions, being able to thermal image a large area and look for warm bodies may be very useful.

Of course I have no idea if this is feasible or not, but I assume someone here will correct me if I'm pipe-dreaming.

its very feasible with satellites and drones run by DARPA or USAF. Even helicopters use IR and "night vision". the sensing systems are expensive
There are commercial sats doing this.
Not live at that resolution, AFAIK.

Planet has the spatial resolution but not the temporal resolution.

NOAA has the temporal resolution but not the spatial resolution.

Please correct me if I'm wrong.

Planet has sub 10-cm resolution? That's surprising
50cm, not 10cm, but enough.

TBH I thought they had 15cm, but for human-scale large object monitoring (fires, traffic, construction) 50cm is enough.

Why does forest fires need 10cm resolution? I'm fairly familiar that space (in Australia though) and I've never heard that requirement.

There are other fields where 10cm resolution is required though, obviously.

Might be great for preventing the spread of small fires that just started. Dump some chemicals on it 10 minutes later and prevent a bigger problem.
Very interesting! This sounds a lot like the more recent iterations of the Airbus Zephyr: https://en.wikipedia.org/wiki/Airbus_Zephyr

I don't know how much you can share, but I'm curious about the solar and battery system.

Are you using silicon cells, multi-junction cells, or something else?

What sort of battery chemistry do you use? It seems like deep battery cycling is a must to keep weight under control, but battery degradation is probably the limiting factor for mission duration. The low temperature in the stratosphere also seems more challenging than what terrestrial vehicles deal with.

You beat me to it! Competition is good, but so far none of the various start-ups trying to build aircraft have anything to show (boom, all eVtol companies...).

Edit: Funny, they didn't mention Airbus anywhere. But they do have the perfect founder background, Amazon Air and delivery drones. Sells good with VCs, but so far they have to get real aircraft in the air.

Their YouTube demo shows a full 24hour flight.

They’re the perfect scrappy startup with fast iteration cycles.

Yes, Airbus’s project and team is very cool! What we’re doing shares similarities with Zephyr, but we’re focusing on a lower cost system, and a larger payload which will enable more use cases.

We’re using conventional silicon solar cells, rather than the GaAs cells used by Airbus. We give up some efficiency by doing this, but it keeps costs far lower - which we think is key to iterating quickly, and opening up some of the market use cases.

Similarly, we’re sticking with conventional battery chemistries (Lithium Ion). Battery energy density is by far the biggest driver for this technology - in the past decade we’ve seen huge advancements in battery tech, which is one of the reasons this technology is now possible. As you identify, cycle life is a key challenge and what will limit the aircraft’s endurance.

> we’re focusing on a lower cost system, and a larger payload which will enable more use cases.

So you want to beat Airbus but also be cheaper. And do it with off-the-shelf components too, because those specialized ones are just too expensive. Not to say it's impossible but wow

That’s also how SpaceX disrupted the launch industry. Rapid iteration of cheap shit lets you learn so much more and improve the system so rapidly that competitors get left in the dust.
the main cost is labor. Airbus probably has >1k people working on that project.
I think this is a bad take - large corporations like Airbus are terrible at R&D. They're simply not optimised for it. It's not very hard to move much, much quicker than a large corporation does - because the people inside the large corporation are having to contend with internal politics, legacy tools, legacy cost structures, and corporate inertia that doesn't like doing things a new way.

If you've worked in both a startup and a large corporation doing R&D, it's really obvious that startups are far, far more effective at it. If you gave $10m to a startup, vs a budget in a corporate project, you'd probably get 10x the result from the startup.

Where corporations excel is in scaling things up. Once you have a good design, you need a good manufacturing process and a solid supply chain. Startups rarely have the capital or knowledge to put that in place, while big engineering corporations will.

Why go with heavier than air?
I'd guess because it's less prone to get interfered with by wind, more easily controllable, and can stay up longer if battery power lasts overnight.

I too would like to see more autonomous hydrogen balloons flying. Even on tethers to get height cheaper than a tower.

I'm not well-read on the technology involved. Is it possible (feasible?) to use a high altitude balloon to get a plane to altitude?

You could either detach from the balloon, or use spare solar power to deflate the balloon in case you need it again. Maybe the balloon could help maintain altitude at night.

Perhaps not as high altitude as the stratosphere, but the US was experimenting with deployment and 'landing' of fixed wing aircraft from airships in the 1930s. https://en.wikipedia.org/wiki/USS_Akron
They also tried building aircraft carriers out of ice in the 40s. Doesn't mean it was a good idea.
Probably for control reasons. You can send balloons up to that altitude, but they'll go where the wind takes them.
Each has merits, but in our evaluation, if you are able to overcome the aircraft design challenges, fixed-wing offers a far more capable platform. The stratosphere has winds that can exceed 100km/h. With balloons, station-keeping is an enormous challenge. We spoke to a lot of people in the Google Loon project, and ultimately this became a defining problem for them. Airships/dirigibles help a bit, but still have to drift in strong winds - beyond that, they also become extremely large and expensive to work with.
What fraction of the power budget do you think will remain for non-propulsion?
(comment deleted)
Very cool idea. As far as applications, I love the idea of this being a lower cost / higher resolution version of satellite imaging, think Ororatec, Planet, Spire etc.
Absolutely, we think there are some really interesting use cases around imagery too - for exactly these reasons.
Very cool. I used to dream of this stuff when I was younger. Reminds me of Atlantik Solar: https://www.atlantiksolar.ethz.ch/. Hasn't been updated in a while, but focused more on low-altitude autonomous survey missions.
Atlantik Solar was a very cool project, read a lot of their research - count me as a fan :)
This sounds like military technology of the kind that e.g. Ukraine would currently like a few samples of. Or at least, dual-use.
Would probably be shot down or hacked/jammed by the Russians in a day or two
Shooting down an airplane at 20km altitude requires an expensive rocket and reveals position of the launcher. No, can't use cheap MANPADs for this.

My gut feeling tells me this airplane is potentially cheaper than a rocket, so if it is shot down, it is a win already. If it detects launcher position while rocket is on the way it is even bigger win. Potentially it reveals larger radar position too, that's another win.

It is difficult, expensive and risky to jamm objects at that altitude. In the current Ukraine-Russia war russians would probably not bother jamming it.

So, what these guys are developing certainly is a military technology. No way around this. High endurance, high altitude cheap aircraft will sell like hot cakes.

I don't know how they communicate with FPV drones 20-30Km away on the battlefield. They can't use cellphone networks, because they're compromised. I doubt you can fit a starlink terminal to a drone.

But this thing could be fitted with frequency-hopping comms, and could control a swarm of attack drones from 70,000 feet. It could also have a backlink to the pilot, allowing him to control any drone in the swarm.

I imagine the Ukrainian nerds are already all over this sort of idea.

The Ukrainians have infact fitted starlink to aerial drones. One of the reasons for their disputes with SpaceX.
This looks really promising! Really stoked to see where you guys take this.
Thanks, next stop is the stratosphere! :)
I was peripherally involved with a similar project a few years ago. They got bought out by Facebook. This was before Aquila.

There’s a good reason why most airplanes look and behave alike: because it works. Technology choices like flying wings (not used here) or twisting wings (used here) are a “code smell” to my computer programmer nose. To me, nothing is sexier than good reliable technology that works. Smart investors will know that. The other ones will do you a favor by staying away.

The use of batteries to maintain altitude at night is a big concern. Batteries are heavy and will dramatically cut down on the usable payload. They need to be maintained warm (but not too warm!), which is a challenge at the low ambient temperatures the aircraft will experience while flying.

Most airplane accidents happen during take-off and landing. Please do not shortchange these phases of flight, and good luck.

To be fair, the Wright Flyer used twisting wings. Ailerons took over, but it's not like twisting wings are some low TRL thing.
The Wright brothers were incredible engineers, and a twisting wing made sense for their design. For high efficiency, a stiff wing is a must. Nowadays active stability systems make more designs practical. But the mission goals here favor low weight and reliability.
Also the Wright bros’ litigiousness over their IP was a main driver of the effective extinction of wing warping and the rise of ailerons rather than strictly engineering concerns. I imagine with newer materials wing warping could be pretty effective.
Low weight and high aerodynamic efficiency are crucial. You naturally end up with high aspect-ratio wings which tend to be more flexible than conventional aircraft wings. In order to keep aero-elastic effects under control, multiple tailplanes can help to stabilize the structure, and provide roll control at the same time - we’re not set on this technology though, ultimately our analysis tooling will drive this design decision.

Batteries are heavy, but if you want to conduct long endurance flights - greater than a few days, then realistically they are the best option.

What are the alternatives to batteries if you want to fly through the night?
Do you have any ethical concerns about the use of your technology for surveillance of civilian populations?

That’s the obvious use case for this technology, right? Cheaper and more easily retargetable than satellites. Longer duration, quieter, and more resilient than manned aircraft, but not fast or maneuverable enough to survive in contested airspace.

So that leaves large area, long term surveillance in uncontested conditions as the prime use case.

Sounds an awful lot like cops or an occupying force putting a fleet of these up to indefinitely track all movement or signals over entire neighborhoods or cities. Do you plan to market and sell your technology for this use?

Excellent question, this sounds like an enabling technology for mass surveillance on a much larger scale. Creepy.
I agree that's the most obvious use case, so I think the fact that the founders consciously avoided mentioning it means they must have ethical concerns, or at least understand others' concerns. I think this is a reasonable question to ask, and I'd be interested to hear whether they plan to sell it to surveillors and just feel guilty about admitting that, or if there's some reason they think this won't work for that application.
Shouldn’t ethical concerns stop the usage of a technology, not its creation? Otherwise we risk never gaining the upside of said technology.
We can’t even stop usage of it. It’s futile. Best case is to also invent ways of disabling surveillance.
This is a concern for us. As with many disruptive technical advancements (e.g. in Nuclear or AI), there are many ways our technology can be used - both good and bad. It's important to us that what we do is ethical - that means supporting and pursuing use cases with huge positive societal value like rural connectivity or wildfire monitoring. With that said, there are definitely ethically-questionable use cases for this technology and I don’t underestimate how difficult it will be to navigate. We certainly have no intention of monitoring civilian populations and haven’t spoken to any police forces about what we’re doing.
I'm glad to hear that, I hope you stick to those principles if the police or military come calling. They have deep pockets and your investors will want their 100x.
Making their plane military-approved would be extremely costly and frankly it probably wouldn't work. There have been a lot of people commenting about the military clearly wanting this and taking over the project, but what makes a plane like this possible is not having the same requirements as military aircraft.

My biggest concern would be weight & power. This plane only works if it stays light and doesn't use a lot of power. If the military wanted this for live video or EO/IR, how would that communicate with ground sensors? KU satcom, UHF? Will it have IFF onboard, will it meet all the no-single-point-of-failure requirements? Everything the military requires will start using up a ton of power and adding a ton of weight. I get people always think these things will be used for evil, but it isn't exactly easy to take something civilian and suddenly ship it off to some USAF squadron.

So YC is getting into defense contracting now?
Fascinating reaction. It's not surprising but it is nonetheless disappointing.
Maybe they should be working on reliable management of multiple HN accounts.
I am commenting on my previous comment as myself. I'm not trying to pretend to be someone else. I don't have multiple HN accounts.
did you just respond to yourself as if a different person?
No, I commented on my own comment as the same person. I'm not surprised but I am disappointed to see the response as downvotes rather than actual engagement with the question and its implications. It's almost like people want to shape the Overton window to avoid this kind of discussion, ie, to put the blinders on themselves and others.

YC's reputation depends on and is justified by its investment portfolio. Its investments are not justified by its reputation.

Just wanted to comment to wish you guys luck. There are some bizarre comments here about use of the aircraft for defense purposes (as if that's a bad thing) or the "ethics" of surveillance, which applies equally to any surveillance technology, there's no reason to single out this platform.

I do wonder about the payload capacity, it seems like it might be quite small given the batteries and the limited performance available from the altitude.

Do you climb during the day and glide at night to save battery? And if so, how much difference does it make?
How does the aircraft get to 70k feet? Is it launched from a separate vehicle?
I'm similarly curious. Designing for a higher stall speed permits smaller wing area, lower drag, and lower weight. The cost is that takeoffs and landings become troublesome.
Could be catapult launched?
Lee waves? Thermals? Cumulonimbus thermals? :)
I don't think one could just "throw" something 70k feet up in the air.
Climbing to altitude is the straightforward part. The transition from zero knots to the stall speed of the aircraft (minimum speed at which it can remain airborne) is the tricky bit. Designing for a lower stall speed necessitates wings which produce higher drag (by being larger) which requires more propulsion, which means bigger batteries and motors. So launching from a catapult or rocket or mothership or whatever means a lighter plane.
Launching by rocket means the plane would have to be pretty rugged to survive it. And that means more weight. You also have the issue of deployment. Folding wings means more weight and more things to go wrong.
Spinlaunch would disgree but here the trouble is getting off the ground; Take-off is the most demanding phase of flight in terms of propulsion demand.
Spinlaunch would crush any aircraft that is light enough to stay aloft indefinitely.
Or balloon launched, parachute landing.
Other options that might work: * Launch from the roof of a vehicle * With a glider winch * Towed by another aircraft * Auxiliary engine / batteries that can be jettisoned & parachuted down
The aircraft is able to climb under its own power. We have a diurnal energy cycle - charging the battery up through the day and deploying battery energy in the night. If we launch in the morning with a full battery, we have a whole day's worth of extra solar power to use to climb up to altitude.

Winds will be a bigger issue than energy when climbing. Up at 20 km (70k ft.) winds are quite calm, but we need to ascend through more turbulent winds as we climb. We’re sizing our MVP around this.

Winds up to 20 km are NOT quite calm. They can max out to 60 m/s at height of 10 km, which is more than 3 times than enough to blow a solar aircraft far-far away. You have to choose proper meteo conditions for climbing and descending and plan the trajectory taking into account those winds to be able to land at a given place. I've been analyzing GFS data a couple of years ago for similar project. The problem is that lack of energy demands to design a really low speed aircraft for low densities that you can find in stratosphere. At 10 km winds are stronger than in stratosphere and density is higer. You can find the results in https://journals.sagepub.com/eprint/GVXWXDABPE6A8PNRGTBU/ful.... The whole aircraft model probably is not like yours and is subject to many conservative estimates, but I am pretty much sure that wind model is accurate in this publication and generalizable to different regions.
I'd assume it's like on Mars, where wind speeds are hellacious but there's just not enough 'air' mass to do any real harm.
I think the comment you responded to was saying at 20km ("up at 20km") it is calm, but below this it is not calm.
I suppose an aircraft like this can climb like gliders do, using the streams of air that move upwards. This requires some planning, and likely a skilled glider pilot to help choose the course.
I'm not an expert but I don't think thermals rise that high.
They could help the first few kilometres, where the air density and thus the drag are highest.
Looks cool! Reminds me of that lady replicant from Bladerunner 2049 watching the protagonist through her glasses while getting a manicure.
Hello 20 ft. and 13 lbs, great if you could use the metric system to describe your aircraft :)

Also, in engineering it's pretty common to use it these days, even in the US.

I LOVE this -- I've nearly pulled the trigger on solar drone builds in the past, this is everything I wanted and more.

Amazing work! Super excited to follow along!

This sounds awesome. I’m glad someone is running with this tech and not handcuffing it to a particular use case. Good luck!