174 comments

[ 3.3 ms ] story [ 216 ms ] thread
Gotta be more to it. They blithly say, reeling it in takes a fraction of the energy it generates when it pulls the line out. But that's nonsense on the face of it.

I'm guessing, it changes it's wind profile, maybe going edge-on or closing up, to make it easy to pull down again. Gotta be something.

I think the kite might have control surfaces, thus would be able to control lift by changing its angle of attack.
Trailing edge control lines, pretty standard. You let them out and the kite will no longer be able to catch the wind and will float down. Keeping some tension on those lines and wenching it in will keep the kite controllable.
Brake lines on a traction kite like that can cause the foil to fully de-depower and fall out of the sky. So you can let the wind pull it up, and then let gravity bring it down.
Only as much nonsense as rowing a boat without taking the paddle out of the water. In that specific example, you're going to be going in circles, but you will be moving.
You can propel a boat forwards with only one oar, without lifting it from the water. It's a useful skill, Kayakers use the same stroke to move sideways, and for stability in rough water. See https://en.wikipedia.org/wiki/Stern_sculling
Exactly, that's why calling this concept nonsense is so weird. It might be a tiny bit counterintuitive at first glance, but there's plenty of examples of it all around.
On its face it sounds silly, until you get experience with wind surfing. It's using the same concept. Figure 8s generate the pulling force. When you stop the figure 8s, and change the angle of the kite, you get minimal pulling force.
> Reeling it in takes a fraction of the energy it generates when it pulls the line out. But that's nonsense on the face of it.

Why is this nonsense on the face of it? This is just how kites work. Do you imagine that kite surfers just overpower the kite with their might and pull it back to the ground when they've had enough?

It’s also how turbines work. The fluid pushing on one side isn’t cancelled out by the drag created on the other.
No. I said I didn't. I said I believed there was more to it.

Others have actually elaborated on the likely mechanism. But sure, mocking people is a kind of post I guess.

>The energy generated by the system while reeling out is greater than the energy consumed to reel the kite back in.

There's something about it that just triggers the "perpetual motion machine" identification heuristic.

On a multi line kite you can control the angle of attack and speed. That gives you the ability to dial the pulling force up and down over a large range.
You're not wrong - they do that.

Have a look at https://thekitepower.com/products/ There's a diagram of the flight path and power generation over the cycles.

The flight path on the way out is much much longer going cross wind in figure 8s. On the pull back in it's a direct path back in, and it looks the kite shifts to point back at the base station more so there's a ton less resistance than on the way out.

I’m happy to see this idea continues to develop. I was worried its failure was foregone when Google shuttered their moonshot Makani in 2020. It seems like this iteration is simpler.
for some reason, this always happens with the google moonshots; They come at it from the wrong angle. Someone comes along with a slightly simplier solution after google scrapes the whole thing as uneconomical or beats them to the punch on using a far more technical solution. Don't seem to have a big "Cost per KWH" KPI's internally nor a sense of urgency.
I wonder how Malta is doing. The News blurbs at their website seem... ok?
Not to be a HN naysayer, but wasn't this tried already with Makani Power?[1] They even made a documentary about it[2]

Compared to traditional wind turbines, this solution trades an enormous amount of complexity, maintenance, etc for small reductions in material cost. We should be churning out traditional wind turbines for cheap instead.

[1]https://x.company/projects/makani/

[2]https://www.youtube.com/watch?v=qd_hEja6bzE

The article poses a different use case, replacing diesel generators for off-grid power, not hooking into the grid to replace turbines.
This is a very different approach. The google approach is making an airplane fly in circles generating continuous power, this slowly lets a kite out then reels it in generating pulsed power that would need a battery to smooth out. This approach can use pretty standard kite technology which is far cheaper than airplane technology.
Use multiple kites in a staggered pattern so one of them is always generating near peak power. That complexity of course offsets all of the gains, but that's fine because it illustrates why this is an idea that just won't fly.
Batteries are fine, the majority of the time it'll be in the pull stage as once depowered (the power is taken out of the kite by releasing the brake lines) you can bring the kite back down pretty quickly. Having two with the cycles staggered would reduce this but you need batteries anyway because you can't rely on demand being flat.

The algorithms to control the kite are pretty straightforward, the complexity to self launch would be a bit tough but it's easy enough for a human to do so there is no need to include that complexity for the use cases this is targeting. This wouldn't be a set and forget solution, not yet.

The kites will wear out relatively quickly but they're cheap to replace.

I consider myself a pretty skeptical / practical Engineer and I don't see any show stoppers, I think it'll fit the niche they're targeting quite nicely.

From both a maintenance, operational cost and engineering complexity point of view I don't see this as viable, but I'm more than happy to be proven wrong by a party that brings it successfully to market. Meanwhile, any place where this kite system would work is one where a regular HAT would work as well and long term (25 year lifespan) total cost of ownership and $/KWh generated will be pretty easy to determine by deploying two systems side-by-side.
The article indicates that it comes with an integrated battery.
which also includes a small amount of solar apparently
It'll make sense in some places where it's harder to build/deploy a 40kW traditional turbine. They're not easy to move and are relatively close to the ground so they are viable in fewer places than kite systems because winds aloft are much higher than those near the ground. You can also use this temporarily in areas where you have periodic power requirements and poor access to the wider electric grid. Think about deploying a lot of these in areas right after hurricanes or other disasters.
Not to be flip, but the fact that Google canceled a project doesn’t say a whole lot about whether that project was viable.
The fact that Google funds a project isn't evidence that it is viable either.
> Not to be a HN naysayer, but wasn't this tried already with Makani Power?[1] They even made a documentary about it[2]

So? These (kitepower and skysails, but there are more) are long standing 'competitors' of Makani, all existing for far longer than 2020 when Makani went down. Just because Google gave up, not everyone else also gave up. I mean, Google also stopped Google code, groups and google+, yet similar products are still triving.

60 square meters of kite surface area is a lot of pull. 5 square meters can be enough to pick up 100kg in the right wind conditions.
A sailboat typically flies about 5 - 8 square meters of material in the sails, pulling 3 - 4 tons of material around the world ..
This doesn't seem right to me, in my experience a 3 - 4 ton sailboat is around 9m long, which has a sail in the 20 - 40 m^2 range.

A 5 - 8 square meter sail is really teenie tiny.

Yeah, that's sailboard territory, or a small dingy with no intention of going fast.
Robotic boat, a large tank, some power-to-storable tech, a power generating turbine beneath the waterline and a huge kite. Just send it out on the oceans, to cruise aimlessly wherever the satellite uplink suggests favorable weather and have it head home when the algorithms assumes that remaining tank capacity might get filled up on the way back. Most attractive post-fossil future ever.

Unfortunately, unattended power-to-anything is very much a far away dream, despite all delusions of being able to do it large scale on Mars. All other technology is basically ready to go, and the "lateral lift" of a fast-moving boat is a perfectly adequate substitute for mooring when it comes to tapping wind on the high seas.

Based on the numbers from the article: 40 kW * 80 s / 500 m is 650 kgf.

So maybe it will work. Still I feel like as soon as you get a moment of calm it will all break. The wind may be very reliable in places but is it ever so reliable that you don't get moments of calm for hours at a time?

When there is a break in the wind, you reel the kite in, creating apparent wind.
You can, but in that phase, you're expending energy without generating any.
The diagram envisions the kite also being connected to a battery, diesel generator, and solar PV installation.

The real question is not if the kite installation or renewable installation is 100% reliable, it's a question of how much does it cut your total diesel consumption. US military loves renewables because it reduces the amount of fuel they have to move.

Yeah I see this as making a lot of sense for armies as a first customer. Highly mobile, reliable energy supply you can set up anywhere (assuming their claims are correct)
I never saw that in the Army, but maybe I was in the wrong kind of unit. I wouldn't count on good continuity between statements of the upper officer corps (political appointees) and the operational realities of the line of business personnel. Just like any big organization.
Operating the kite isnt easy on its own. A lot of complexity involved in this. I am intrigued as someone who kites and develops clean energy. I have a tough time seeing how this is a set and forget situation. Maybes it the first step in a direction with something better down the road!
> After the tether reaches its maximum length, the ground station winches the kite back in. Though the Hawk must expend energy for reel-in, it only expends a fraction of the energy, resulting in a net energy gain that varies by wind speed. An entire cycle takes about 100 seconds: 80 for reel-out and 20 for reel-in.

This is an interesting approach. Also, the kite is doing eights all the time. Although I suspect that is only while reeling out.

Presumably they can adjust the attitude to make it easier to real in.
My recollection from flying kites a long time ago is that operating a kite gets easy after a certain height. I imagine the Kite Control Unit (KCU) can easily keep it aloft once it's high enough. The tether is 352m long according to https://thekitepower.com/the-hawk/#components
I don’t think you can make a kite that can last a decade of load. We just don’t have materials strong enough. Also you need a sophisticated algorithm to fly it and never crash
As someone who regularly flies hobby kites, I think it can be done.

Modern kites are made of nylon and kevlar. With regular maintenance these kites should last at least a decade. Cables might have to be replaced every few years.

As for the flying algorithm, I doubt it has to be very sophisticated at all, especially if the kite is itself designed to self right and stabilise. After all, a single line kite has 0 controls.

I sort of assumed it would power and depower like a kiteboarding kite to create work. You could probably depower by collapsing the kite leading edge and dragging it back in but that seems less efficient than flying it in a figure 8 to power and depower.
How is the kite raised initially and what happens when air stops, how is it re-raised?

Reading the article seems like it's pulled back if air is stopping, and slowly released based on air speed perhaps.

On islands where this is targeted the winds are usually very predictable, especially at higher altitudes. This isn't like flying a kite at ground level, there is no need to wench the kite back down to ground level, just up and down where the faster high altitude winds are.
All of the raising is done by the wind.
I wonder how they will deal with bird attacks. Falcons and eagles (ironically, they share the same name with this kite) are territorial and will go after these drones. I once had a huge heavy-lift drone that got attacked too. But with a pilot nearby, the easy solution is to fly to a higher altitude, that seems to determine who rules the sky in their eyes apparently. I’m not sure how this kite will manage such situation.
By being 60m2 big? I mean, helicopters are also not being attacked by eagles, why would this kite be?
Helicopters do get attacked, not so common because of several factors like the size, flight behavior, altitude, and most importantly the engine size and placement, these birds from my observation they usually attack from behind and/or above, both are properly “protected” in the helicopters, not so much in this kite, additionally, when the pilot is flying, they usually avoid it and continue their path, the birds mostly won’t chase and lose interest, but this kite is stationary and if the bird decided that it is hostile, it will keep trying to prey on it especially when you can’t increase the altitude substantially like a normal drone.
Wind power machine idea #1: turbines.

Potential concern with idea #1: the machines may kill birds.

Wind power machine idea #2: kites.

Potential concern with idea #2: birds may kill the machines.

The bright side is that wind doesn't tend to stop as much 325m up
People cannot be anywhere in the radius of the kite cable. Kinda limits where is can be deployed. But the quick deployment could work for some situations.
This is not correct. Untrained people can't be inside the flight zone while operating. This is roughly the downwind quarter of the potential flight zone.
Even a quarter is a huge area. Using the "safety buffer" radius of 425m, that is 560k m2 or 140 acres. A quarter of that is 35 acres, for 30 kW.

For regular wind turbines, it's 30-70 acres per MW generated. That is 20x-40x more power for the same land area. And almost all the land under a wind turbine is freely usable.

> Untrained people can't be inside the flight zone

What kind of training makes you immune to being sliced in half by a tether?

The same kind that makes you immune to crashing a car, none. However the training presumably could help you asses and mitigate risks better than someone with no training.
Are these less visible than traditional wind mills? If so, maybe they could be used at sea and get around the hatred of wind power as an eye sore.
You can probably deploy it inside the solar farm, most of the times these aren’t open for public to wander around.
So it's just a really big battery.

The kite thing is just a novelty/marketing move. There have been containerized commercial battery systems available for at least a decade. The barrier to adoption is, as always, price, not "kite delivery".

Is this for really remote places, like Greenland/Alaska? Because otherwise you can just use a truck/tractor.

You completely misunderstood. The kite is not for "delivery", it actually generates the power.
Yes, you're right. I'll have to re-read the article to see how much this kite can actually generate.
How will this work with aviation? I know it probably won't affect most IFR flights, but still possibly dangerous for VFR. Will there need to be a new restricted area for each deployment?
Depending on the altitude, they could keep these within class G airspace, which means VFR's job is to see and avoid. I think that's a pretty defensible solution as a pilot who spends all of his time in VFR operating in Class E/G airspace.
A project I did a couple years ago was to have an onboard SDR that communicates automatically with airplanes in the vicinity, or switching to manual where the drone operator can communicate as if they are “on-board” through the internet by a mic on ground. I can see something like this is doable, with a modified system to fit the length flight, after all, the position is fixed so I don’t think it would be a problem.
Just wondering, how were you communicating to the airplanes in vicinity?

Radio would require the planes to be listening on a specific frequency and ADSB Out would require the planes to have ADSB In which is not guaranteed.

It wasn’t an ADS-B, Transport Canada (the FAA equivalent in Canada) doesn’t like ADS-B on drones yet, so the solution was to have an SDR (BladeRF, full-duplex for Tx/Rx), the on-board SBC had a server the received the voice sample (either direct through ground station MIC or automated reading directions, alt, etc every X period of time) and then broadcasting it to the airband, so it’s simply:

Ground station mic -> internet -> server on SBC -> SDR -> airband (AM, I think it was ~120Mhz that time) -> other pilots

If you have the manual communication (where the drone pilot comms and not an automated broadcast), you can pretty much talk with pilots as if you are on-board.

I had a better write up if interested on how it works in here, in the “SDR” section.

https://tamim.io/professional_projects/nerds-heavy-lift-dron...

At first look, it seems like simply by having so much less mass of material involved that you have a good head start at competing on LCOE, even if the system needs the kite and tether replaced every 5-10 years.

I think the biggest roadblock would be making sure it can return to the perch with high reliability otherwise you will have a lot of service calls for kites that landed on the ground.

There's been a recent video by the German "Sendung mit der Maus" (basically how does stuff work for little children) about these kind of generators. I did not double check if the company in the video is the German one mentioned in the article, but chances are high. Video and subs are in German only, but it's less than 10 minutes and one can see a little how it works.

https://m.youtube.com/watch?v=aqfFZDWiJOI

Why are these people wearing safety helmets? Does this imply that everyone living under the kites will have to wear one too?
They're under a crane lifting stuff up overhead on a work site, that's probably why.
but it's not a construction site though
Falling bolts and tools don’t really care if they’re on a construction site or not
So let me quote my original question

>Does this imply that everyone living under the kites will have to wear one too?

Maybe just the ones living under the crane while it's lifting stuff over them.

That also assumes people are living under the kites.

I think you're overthinking a short demo video and hats. I doubt anyone fought hard against wearing it, and company liability is different from your own personal risk assessment.

no, it's because the are directly at the site where the kite is launched (and on national tv)
I doubt people will be living under the kites. I doubt there can be anything in the vicinity of the kite flying range for that matter, so no windmills or high tension lines or buildings or anything, which limits this thing's options especially in higher density countries like NL.
Very cool. I noticed that the radio-controlled unit that steers the kite has its own little wind turbine for power.
I wish if there’s an actual video of the whole system being operated and not just 3D renders, a lot of questions in mind about some details but probably most of them will be answered in such video.
I can only imagine how many 3D models they render in a physics simulation platform like Nvidia Omniverse so they likely just used what they already had.
Interestingly, the very first statement is quite misleading. The statement:

> On average, a humble wind turbine uses less land area per megawatt-hour than almost any other power source.

The article itself refers to this citation [1]. It says, unsurprisingly, that the most land efficient power source is nuclear. The statement is of course technically true because it says "almost any power source".

With wind, however, even that is tricky because you can either count just the area directly below the turbine body, which is still higher than nuclear and unreasonably optimistic because you severely limit a far bigger area. If you count also the spacing (too pessimistic on the other hand because part of the land might be used for farming) then wind farms range across the whole scale of land efficiency.

I think its fair to say that land use of wind farms is "complicated", but saying its almost the best in this regard sounds like manipulation.

[1] https://ourworldindata.org/land-use-per-energy-source

Wind turbines can be built big enough that the blades, at the lowest point, are higher than almost all other land uses, except perhaps forestry.

To be honest, the access roads are probably the biggest real impediment to using the land around a wind turbine. Can't exactly farm millions of acres of corn with robotic combine harvesters with a criss-cross of access roads to the base of every turbine.

Why not? Seems easy to teach the robots to not harvest on the road
Odd shapes and constantly starting/stopping the plougher/seeder/harvester mean you get less ROI on your equipment. Farming is super tight margins - if your machines waste 20% of the time starting and stopping every time they pass over a road, your farms profits are wiped out.

This is true whether the machines are robot driven or not, and it's the reason most commercial farms have huuuuuge fields rather than lots of small ones.

Why would the access roads create a complex shape? Fields already have access roads. Either the farm is flat and the turbine access roads can be on a minimally invasive grid or the turbines are on a hill and the fields are already a complex shape.
(comment deleted)
The cost of the land to the grower is lower on account of the presence of the turbines.
I have trouble imagining that the cost of interrupting your seeder or lifting your plough every quarter mile is really higher than the profit generated by the wind turbines.
That's a turbine in the middle of active farmfield, off a dirt road, servicing a town a bit under a mile away that is barely 10 city blocks wide at its widest point. In the middle of Iowa.

My point just because you can build the turbines to clear everything, there are areas in this nation where there's just no point to doing that; and it's more economical to just make them shorter and put the money towards something more useful

OK, I think I agree but I was responding to london. The idea that wind turbines and large-scale corn cultivation are incompatible is contrary to an abundance of counterexamples.
You want your wind turbines as high as possible anyways. Higher up there are stronger and more consistent winds.
Midwestern farmers very much disagree with that “impossibility”
I'm not sure robotic combine harvesters are a real concern (yet ?) - modern "manned" harvesters are already crazy effective and fast & I've seen them navigate some pretty crazy field geometries at times, so I'm sure they can handle a straight road.
> Can't exactly farm millions of acres of corn with robotic combine harvesters with a criss-cross of access roads to the base of every turbine.

You certainly can; single-objective suboptimality is not the same as impossibility. But perhaps this would be an excellent use-case for heavy-lift dirigibles.

Sounds like the article used just the right wording then.

“…less land area per megawatt-hour than almost any other power source.” As I read it this implies there is one (or some very small percentage) power source which uses less land per megawatt-hour, which is exactly the case.

Tidal energy.
An offshore wind farm by definition uses no land area :V
Aircraft carriers suddenly looking competitive...
I'm wondering if the OWID data takes the 3,400km^2 of exclusion zone from Fukushima and Chernobyl (more like 5,000km^2, see below) into account when calculating the average size of a nuclear plant?

I think if we're saying the space beneath the blades can count towards aggregate wind-power land use, an argument can be made to include nuclear wasteland in the aggregate nuclear-power land use.

Chernobyl's exclusion zone is 4143km^2, and Fukushima's is 807km^2.
Shouldn't you include in the average nuclear sites that didn't explode?
(comment deleted)
Why? Nuclear plants can and do melt down.
There are nuclear plants that have been decommissioned and have not exploded. So using just this data, there is a probability that a given nuclear plant will melt down in its lifecycle. And that probability is greater than 0, but far less than 1.
Yes but wind turbines can’t melt down in that way. Even in a worst case scenario the literal blast radius is smaller. This has to factor in to the possible locations.
Of course it doesn't. We don't build nuclear generators with pre-approved unusable exclusions zones. This is a ridiculous suggestion.

Should we consider the potential blast radius of the catastrophic failure of wind turbines as well?

You literally cannot build structures in the path of a wind turbine if you want it to function.

>Of course it doesn't. We don't build nuclear generators with pre-approved unusable exclusions zones. This is a ridiculous suggestion.

I'm not suggesting we assign 5,000km^2 to each individual nuclear plant, just in case they explode. I'm suggesting the 5,000km^2 that currently exists as barren nuclear wasteland should be included in the total land attributed to nuclear energy. That would bring down the average generated energy per km^2 pretty significantly.

The dataset does what I've suggested for hydroelectric. The dams are relatively tiny (and in fact, would outperform both wind and nuclear in this dataset), but they render huge areas unusable. The unusable collateral land is attributed to hydroelectric power, as it should be.

It stands to reason that collateral damage caused by nuclear should be included, if collateral damage by hydroelectric is too.

> I'm suggesting the 5,000km^2 that currently exists as barren nuclear wasteland

Aren't you exaggerating by quite a bit?

It appears it's not far from the truth. According to https://www.britannica.com/story/nuclear-exclusion-zones it's 4100 sq.km. for Chernobyl and 371 sq.km. for Fukushima (800 sq.km. at peak) which isn't an exaggeration. Interesting.
i guess we think of different things when we read "barren nuclear wasteland".
I see. You weren't saying the size was exaggerated. You were saying that the unusable land was being exaggeratedly described as "barren nuclear wasteland". It's unusable land, so it is nuclear wasteland, but it isn't barren AFAIK.
At least we can put lots of nuclear plants in that 5,000 km^2 area so that definition becomes irrelevant for this discussion. Maybe putting the absolute maximum power generated if we turn all available lands to that purpose only is the best metric
> I'm suggesting the 5,000km^2 that currently exists as barren nuclear wasteland should be included in the total land attributed to nuclear energy

It's definitely not 'barren'. I'd suggest it should be counted as nature reserves. See:

"How Chernobyl has become an unexpected haven for wildlife" https://www.unep.org/news-and-stories/story/how-chernobyl-ha...

> Should we consider the potential blast radius of the catastrophic failure of wind turbines as well?

Please? (Not as policy, just, like, in general.)

I imagine falling over and/or yeeting a blade-tip is about the worst thing they're going to do, right? I've know they can burn, but I don't think they explode. What's literally the worst that could happen?

> It says, unsurprisingly, that the most land efficient power source is nuclear

Not true if you factor in waste storage and the time to store it

Long after all the nuclear energy has been used making your toast, and long after many cycles of climate change, you will still be storing hi level nuclear waste, and monitoring it.

But who cares about tomorrow? Party on, and let future generations pay!

How does this calculation work?

Take simple chemical pollution, for example with lead - as far as I can tell, it stays around forever and we,ve poisoned vast landscapes with it during mining coal copper for example.

Or take simple industrial waste , say wind turbine blades, they go to landfill, sure they aren’t hurting anyone but there are a lot of them- does it count?

Other causes of pollution (related to lead, coal, copper...) aren't excuses for nuclear pollution.

AFAIK more and more (many even most, nowadays, as it is legally mandatory in many European nations) are recycled.

looking at the article, you're really splitting hairs - wind and nuclear are 0.3 and 0.4km2, many of the others are 20+

if you were to ask someone based on that if wind was one of the techs that use least land, it's an obvious yes.

I think you're being more disingenuous than the article..

why not a blimp?

Come to think of it, I'm kind of curious why a RC blimps have never become popular. I own DJI drones. But no blimps.

Helium supply is an issue and blimps are risky in adverse weather. Also these kite models keep the generator on the ground, saves from trying to float so much delicate weight.
Interesting! I always try to guess how these things work before I read the article. My imaginary design would have the kite pull a heavy weight up and as the weight moves, it rotates a generator. When the weight reaches the end of the travel, it falls back down and the cycle repeats. The flaw with this design is if the wind was strong enough to pull the weight up, then the weight is not going to be able to fall back under its own power. That's why the real design just uses a motor to do that. Makes sense!
I used to work on this at a different company, and I'm glad to see work on it continuing! It's one of those things where the physics works out really great for fantastic power generation, but the engineering is just so damned complex, so it's tough to... get off the ground.
I think the Google approach suffered from too much engineering. Specifically they had sunk too much money and too many people at very expensive prototypes before they'd gotten the design even close to running reliably. As soon as you become a big team, you become less nimble - which isn't what you need when your product still has big unknowns.

They should have started with five 2 guy teams, a year and a budget of $50k each, and tried 50 different ways to get a scale model that could reliably not crash in all weather conditions for a few months on end.

Only when you have a craft that can either survive all weather, or reliably pack itself away whenever the weather changes quickly, then figure out how to scale it up and make it generate power.

The back-of-the-envelope analysis for kite-based power is very compelling; enough that I tried my hand at designing a kite-powered system in ~2004. I don't think anyone can really appreciate how challenging the engineering (control systems) are for all-weather kite-flying until they just try to reel-in & reel-out a kite on a relatively calm day. I literally couldn't even write-down the differential equation which describes the inverse control path for the kite, when there's one vector for the kite, and one vector for the tether, under the reel-in load; none-the-less, solve the equation, and then get it to work with a freakin' stepper motor.
Don't take the hard maths approach when you can take the lazy AI approach... Just fly a kite by hand, log the data, stick it through a reinforcement learning system to make you a controller that keeps the kite in the air.

Reinforcement learning loves these kind of problems that only have double digit numbers of scalar inputs and outputs.

Either way, my kite flying experience tells me that your reel-motors must move with quite some speed and power if you are to keep the kite in the air when the wind suddenly reverses direction.

I wonder if it would work better if there was a swarm of kites - or at least several deployed in the same area. Far enough from each other to avoid interference, but close enough so they could share data about the air movement.

I agree with the sentiment that going at it from first principles sounds like a nightmare; teaching-by-example AI approach seems like a faster path to success.

I wonder if you could make the kite itself lighter than air to avoid issues where the wind dies down? E.g. a kite shaped airship.
There are no issues, at least as long as you can supply power to the base station: just make the motor/generator reel in the kite, that causes plenty of movement relative to the stationary air to keep it afloat and controllable right down to the reefing mechanism.
Is there anything stopping them from adding a turbine system to the kite that would continuously spin (when there's wind ofc) instead of reeling it in/out?
That increases costs because now you need transmission lines down to the kite instead of only the dyneema lines for control.
Why would they ratchet the kite in instead of making it dive and quickly winding the cable in? Should be easy enough with software right?
Stalling is more controllable than diving
How long does the Kite and Dyneema rope last before either breaks and one or both need replacing? Their site https://thekitepower.com/the-hawk/#space_requirements mentions the lifetime for the battery and ground station, but I'd be most curious about the durability of the whole system.
I suspect looking to sailing would get you the best answers as an outsider. Dyneema ropes are used for sailing but I don't know about the replacement rates, or if they are replaced preemptively for racing where the risk of an issue after some use is too expensive and replacement is cheaper. I have have seen lots of parachute or hot air balloon fabric offered because companies have a limited number of hours of use before they retire them.
After seeing Big Hero 6, I've always wondered about the feasibility of giant floating blimps with generators...
As a longtime kiter and sailor, the durability of the kite and line system seems to be the biggest weakness. Constant uv exposure and stress to these systems would make the economical calculations interesting. I don't know of a current setup that could see sustained continuous use for 1 year without significant material replacements or maintenance...