A reasonable guess would be that their key innovation is actually just lowering the cost of the heliostat system with more complicated software. Traditional heliostats do exactly what you suggest, but it takes a lot of expensive installation, calibration, stiff mechanical structures, precision motors, etc. to make the real world behave like the math. If I had to guess, they are instead mounting a camera at the focal point (plus probably one to each side) and using pretty standard computer vision techniques to measure whether each mirror is pointed correctly. Translating "pretty standard computer vision" into "AI" is just a play for investors and headlines, but taking advantage of a couple cheap cameras to measure and control hundreds of mirrors is a novel and significant advancement.
I guess having a closed control loop makes it way easier to set up the mirrors. You don’t have to know precisely where they are (location and orientation), and don’t have to keep checking for ground subsidence.
The article mentions seamlessly switching to solar when the sun comes out, and hints at having more precise temperature control. I wouldn’t know whether either makes much of a difference.
That's kind of neat. Just four cameras near the target point, watching the spillover from each mirror of the mirror array, and adjusting each mirror.
Keeping the cameras from being blinded by direct hits during startup and adjustment might be a problem. They probably have to start with all mirrors off-target and bring them onto target one at a time.
At a glance they appear to be trying to do solar-thermal chemistry to generate fossil-fuel free cement directly[1], rather than solar power generation via e.g. molten salt. I'm not familiar enough with the space to weigh in on utility/difficulty of their approach.
[1] Probably normally natural gas fired burners/kilns
Concentrated solar power (CSP) is used for energy production with mediocre (though improving) performance/price, and heats to like 500°C. As the article states their technology doubles the temperature and is cheap to deploy, applying AI in the process to do so.
Sounds like a huge step forward to me. And Bill Gates is usually on track with his investments in transformative tech..
Hydrosol-II is a thermo-chemical solar power generator (https://cordis.europa.eu/project/rcn/75098/factsheet/en). It uses solar energy in combination with recycled outputs from process to reach the desired temperature. It is not capable of reaching the desired temperatures (1000 C) purely through solar input.
Heliogen claims to reach the desired temperatures purely through solar.
I'm trying think of anything weird that ML may be able to pick up like correcting for local atmospheric conditions, more even heating over the surface of the collector target, adaptive calibration for better aiming or ability to use cheaper components, etc.
Bill Gross, Heliogen’s current CEO previously founded eSolar. eSolar proposed using an array of Heliostaticly controlled mirrors to direct light toward a solar cell:
So... I’d guess that part of the eSolar approach is carrying over into Heliogen. The control system for using mirrors seems potentially more complex. I assume you want to direct light from an array of mirrors such that it optimally a single cell. Or so that it covers a number of cells, or such that it optimally avoids defective or low performing cells.
So with, some skepticism, I can kind of see an AI play here.
I think the mention of computer vision is spurious. Their website doesn't mention it, or anything related. Any modern heliostat CSP system would use computer guidance, based on the predictable trajectory of the sun. There's nothing revolutionary here; this is just hype.
You're right that they don't mention computer vision. However they do say
> The breakthrough in Heliogen’s technology starts with our patented closed-loop control system that makes our field of mirrors act as a multi-acre magnifying glass to concentrate sunlight.The HelioMax system is an industry first and a critical step in harnessing the power of the sun. Our ability to concentrate and capture sunlight allows us to create carbon-free, ultra-high temperature heat (HelioHeat) commercially for the first time....
They define the HelioMax system as
> HelioMax: An array of computer-controlled mirrors (heliostats) collects and concentrates sunlight.
So, obviously they are trying to not say anything useful about what their special sauce is. But they do seem to be claiming that it has something to do with how the mirrors are controlled. As you say, any modern system will use computer guidance, and the sun's trajectory is very simple and predictable, so it's hard to imagine what exactly they are doing special.
Maybe some sort of active correction for atmospheric distortion? I dunno.
One of the articles I saw mentioned that this is cheaper than other systems due to not needing the mirror mounting systems to be as rigid. The impression I got from that article is that other systems are not very tolerant of any flexing in the structure, which messes up alignment and so reduces output.
This fits in with something I saw in another article, which said that Heliogen's system doesn't need months of calibration when installed before it can produce maximal output.
This suggests that they have some way of measuring the contribution of each individual mirror to the total output, and tweaking its position to maximize that, so that all they have to do is get each mirror near the right position and then they can quickly tune it in dynamically.
This reminds me of a film that the professor in APh 23, "Demonstration Lectures in Optics", showed us at Caltech in the early '80s. The professor [1] was also a researcher at Hughes, and the film showed a demo of a system they were working on there.
It consisted of a bunch of radiators. By adjusting the phase of the radiators to change the interference pattern they could get it so the radiation pattern had a strong lobe in one direction, which they could steer. In theory, all you had to do was crank up the power, and you've got yourself an energy beam weapon that will zap whatever you aim the lobe at.
But how to aim that lobe? That was the cool part. They would modulate the phase of each radiator, with each one having its own unique modulation frequency. Suppose now you've got a target flying around somewhere in front of the thing. It's getting hit by all the radiators, but isn't in the lobe. You have a sensor that can see the energy reflected off the target.
You analyze that reflected signal and look at the frequency components of its intensity variation. If a given radiator is at a phase that is trying to put the target in the lobe, you won't see much variation at that radiators phase modulation frequency. If a given radiator is at a phase that is trying to put the target in maximum destructive interference instead of maximum constructive interference, you will see its frequency in the reflection variation. You can use this to derive a feedback signal for each radiator to adjust its phase to try to make constructive interference at the target.
In the film it showed this system aiming against a dark curtain with no target. You could see a bunch of blobs of light just kind of drifting around aimlessly. Then they dangled a little aluminum model of the starship Enterprise in there, and pretty much instantly all the blobs of light from all the radiators converged on it.
Note that since the feedback is based on the contribution of each radiator at the target, it should be able to automatically compensate for atmospheric distortion.
I wonder if something kind of like this could be what Heliogen is doing? It wouldn't be able to be quite the same, because there wouldn't really be any noticeable constructive or destructive interference going on between the reflections from the different mirrors. But the general idea of modulating each radiator (mirror in this case) and then detecting that modulation at the target to provide feedback for controlling the radiator might work.
If you either had something that could detect small changes in heat over very short times, or small changes in light over very short times, you could put that at the target (maybe more than one, both in the target area and just outside it). Then modulate the mirrors, and look for corresponding variation at the target to figure out if the mirror is positioned right or not.
You probably would not need to do all the mirrors at once, which simplifies things. You could just scan through them, for each modulating it, determining the correction, applying it, and moving on to the next mirror.
If you have good enough computer vision, maybe that part about detecting small changes in light over very short times could be done by a camera looking at the tower.
Cement markets are pretty local. It's cheap enough to manufacture and the raw materials are so abundant that it doesn't make sense to transport cement long distances.
If this were a big enough cost advantage, suddenly it might make sense to transport over longer distances. Putting the carbon saved back into the process. The law of unintended consequences is a harsh mistress.
Cement is too heavy to transport economically over long distances. I remember reading years ago that something like 90% of cement is made within 35 miles of where it's used.
In the long term it may be cheaper once ground transportation is fully electrified and air is running on carbon-neutral synthetic fuels. During the transition period I would expect transportation to be more expensive.
Why? Won't the advent of carbon-neutral synthetic fuels made with abundant solar energy (whether PV or thermal) start lowering the price of ICE-driven transport? What likely events would happen in the interim to raise the prices higher?
I predict that a carbon tax (or something similar) would motivate replacement of all ICE ground vehicles with battery electric before carbon capture becomes cheap. Then later when synthetic fuels arrive they'll only be used by airplanes. It's possible technologies could arrive in a different order though.
Oh, yeah, a carbon tax would certainly make it more expensive to run ICE ground vehicles. I don't think of that as a "technology" though.
When you say, "when synthetic fuels arrive", do you mean the Fischer-Tropsch process developed during World War I? I think that's the most likely candidate process for synfuels. It's reasonably efficient but it uses a lot of energy, because fuels contain a lot of energy — that's why we use them. By coincidence last week I read a bunch of papers by Heather Willauer et al. of NRL about producing synfuels with Fischer-Tropsch from seawater, using an electrolytic acidification process with cation exchange to drive CO₂ out of the water and generate the requisite hydrogen at the same time. The paper with the most comprehensive workout of the costs was ECONOMIC COMPARISONS OF LITTORAL PRODUCTION OF LOW CARBON FUEL FROM NON-FOSSIL ENERGY SOURCES AND SEAWATER, from 2017. You should read it if you're interested in the topic!
So they’ll use something else while California, one of the largest states in the US across any measure, will save a lot of carbon with this route. Hurray, progress.
I think the challenge is that when the sun is variable, you have to constantly regulate the manufacturing process to get consistent output. You have to see clouds coming and scale back the feed rate of whatever material you're heating in advance.
They might have but I can't see it. It looks like they are splitting water to make H2. Quick search showed that there is a company in Spain that had done so since 2008.
You're referring to Hydrosol, an EU-funded prototype thermochemical solar reactor which produces hydrogen by splitting water (and using some of that resultant hydrogen as fuel to scale up the process to industrial temperatures).
That is not at all what Heliogen does. Both products are similar in the sense that they use concentrated energy, but Heliogen has more in common with concentrated solar power plants (like the one between LA and Vegas) than it does with Hydrosol.
I get frustrated by the wide use of the term AI. What is AI other than, well, programming?
Even when I've spoken to devs who have have made game AI's or conversational AI's it seems that a lot of their work is with good 'ol IF statements and exceptions.
Somehow I had always imagined that an AI would be more like a neural net with an operational engine that didn't require exception based programming.
I agree with you more or less, my idea of AI is a program that when given input will not only respond but observe it's response and continue to modify it's response to find a unique (and optimal) solution. I think that this necessitates self-modifying programs but I am no expert.
I also think a lot of the trouble I have with the term stems from the ambiguity in the word artificial. Artificial smile vs artificial diamond.
So in principle, couldn’t you aim the mirrors in a straightforward way by attaching some kind of radiation emitter (maybe pulsed) to the collector? If the mirror can sense the sunlight angle and the incident angle of the sentry, you don’t need any additional position information — the mirror plane is normal to the angle bisector.
I still don't understand the virtue of being a "serial entrepreneur", especially with no indication of what businesses were created. It just sounds like someone who bailed on several previous companies that aren't even worth mentioning.
> I still don't understand the virtue of being a "serial entrepreneur", especially with no indication of what businesses were created.
They certainly should have elaborated a bit. It's Bill Gross from Idealab. He has a long entrepreneurial history and quite mixed record of meaningful successes and spectacular failures.
"Prior to Idealab, Gross founded GNP Loudspeakers (now GNP Audio Video), an audio equipment manufacturer; GNP Development Inc., acquired by Lotus Software; and Knowledge Adventure, an educational software company, later acquired by Cendant."
Write more details about the tech, potential hurdles or challenges, and how they might be addressed. Don't mention or quote investors (literally people with vested interests).
Now that I think of it, the idea of the wiki-tribune crowd-sourced news and social network is appealing, and has many similarities to HN.
This is really nice. I just wonder whether it heats the air so much near the focal point that it creates an updraft that pulls in and fries insects and birds? (Not that this would stop it from bring a net-positive.)
Yes, but the Odeillo target is 40 cm in size...which is fundamentally too small for industrial-scale heating.
Heliogen isn't getting anywhere near that temperature...but it is apparently getting to industrially-useful temperatures at industrially-useful sized targets at financially-feasible costs.
You also tend to have higher air pressure in hot sunny area's needed for this tech and that will also increase the melting point of iron. So whilst a small detail, certain one that would be measurable on running costs/energy needed.
Someone was asking me about trigering fusion with this, and i think fusion at the core of the sun is at millions °C, and at the surface you get 5500°C, so beacuse photons carry entropy and the photons of the surface is what you see, you will never be hotter than the surface of the sun, even as much rays you can concentrate. It is not a mather of how much rays you get but of pure entropy. Hotter means more entropy. You can not create more entropy in this case.
Photons don't carry entropy, creating more entropy is easy, and hotter doesn't mean more entropy. However, you're correct that you can't reach temperatures higher than the surface of the sun by concentrating sunlight (because that would destroy some entropy), and that you need millions of degrees for an appreciable rate of fusion.
Solar furnaces https://en.wikipedia.org/wiki/Solar_furnace have been useful for reaching especially high temperatures since at least Trombe's 1949 furnace, which can hit 3500°, without computer vision or even closed-loop control. So why is a 1500° solar furnace being touted as a groundbreaking innovation and a new high-temperature landmark?
A friend asked me what I thought about this earlier, based on a somewhat better reprinting of their press release:
Scroll back to 2010, when Bill Gross started working on this. That's when he got funded by that dude whose futurism book about the Information Superhighway, The Road Ahead, didn't mention the internet. In 1995. In 2010, photovoltaic modules cost €1.62 per watt. Concentrating solar power was a promising alternative; it uses the same steam engines used by coal and nuclear power plants, so at scale it should be just as cheap as they are, as long as you can get the cost of the heliostats under control somehow and scale up. It also didn't have that pesky intermittency problem PV modules have: you can store the heat overnight.
Since then, though, heat engines have become economically uncompetitive relative to PV, because PV modules now cost €0.19 per watt, where they've been stuck all year. And steam turbines, almost a century and a half old, aren't improving or getting cheaper rapidly the way PV has been. Being just as cheap as coal isn't a blessing anymore; it's a handicap.
So, if you've been working on CSP and filing patents for a decade before getting your pilot plant up and running, a decade during which the PV market has left your product's price in the dust, what do you do? You look for a possible use where CSP is still viable, such as process heat; you hire a good PR firm; you announce that you won't be building any plants, but you're "willing to partner with" companies that want to build your design; and you hope to God nobody says "Solyndra".
But what's the actual invention? It seems like the actual news is that Bill Gross has patented some aspect of his closed-loop control system using webcams and GPU-accelerated CV to figure out where the mirrors are pointing to improve your concentration factor. The key improvement that made it possible was better GPUs, according to the press release, anyway.
So what happened, from the point of view of anyone outside Idealab, is that now Idealab and Intellectual Ventures will sue you if you do this fairly obvious thing of using high-resolution webcams for precise heliostat control.
So, when would this be a sensible thing to do?
Trombe's solar furnace and similar devices are able to compete quite effectively in the "market" for process heat at the 2500–3500° level, since, as I said, 1949. (I guess Bill's PR firm didn't know this, or hopes you don't.) That's a level almost impossible to achieve using fire (oxy-acetylene burns at 3500° under ideal conditions), and difficult even with arc furnaces. But Bill's thing is designed for a more prosaic 1000–1500° level, where it's competing not only with fire but also Kanthal or SiC fed from PV, wind, hydro, and nukes, as well as induction, dielectric heating, and microwave heating.
The potential advantage of CSP for process heat at these lower temperatures is that it's cheap and abundant. If you fill a field with mirrors, they can harvest 6× as much power than PV modules covering the same field can. But if land area is your limiting factor, your cement plant or steel mill or whatever probably isn't in the middle of a big field; it's using a lot more energy than your land receives in sunlight. In that case, you probably want to pull your power from someplace further off, whether in the form of coal, oil, gas, biomass, or electricity. Probably electricity from PV panels if we're talking about anything post-2030.
But suppose you can put your factory in the field where the mirrors are, and the limiting resource isn't land but money. In that case, it might be a reasonable approach. PV modules cost €30 a square meter now. That's probably more expensive than mirrors, if you take into account that mirrors give you 6x as much energy: €180 per square meter is the price mirrors have to beat, and that seems doable.
But now you are on notice: if you do that, make sure it's in a country where Intellectual Ventures's shell companies haven't gotten a patent on it, or you have to deal with patent trolls. The press release reprinted above is clear: as with IV's laser mosquito swatter, they aren't going to make it happen themselves, but they'll definitely "partner with" you if you try.
I think we're about to see a giant boom in shitty "do well-known thing X, but with computer vision" patents similar to the shitty "do well-known thing X, but on a computer/on the internet" patents that plagued us in the early 2000s. The availability of massive GPU power means that many things that used to be impractical to do with video data have become possible.
The wikipedia article you linked doesn't actually reference any supposed solar furnaces capable of reaching or exceeding 1000 celsius. Even the project linked in the 1000 celsius bullet point states it only reaches approximately 500 celsius on its own wikipedia page (and at the project's own website).
Note, importantly, as with all things scale matters. There are solar furnaces the size of a pot that have reached 3000 celsius or more. This is not very useful for industrial scale activities.
Presumably one of the breakthroughs is that Heliogen can achieve 1000 celsius at an industrial scale which is not something that appears to have been done before with purely solar energy.
> The wikipedia[sic] article you linked doesn't actually reference any supposed[sic] solar furnaces capable of reaching or exceeding 1000 celsius[sic].
It does; it references the one at Odeillo, which exceeds 3000°, according to this open-access paper by one of its instigators: https://journals.openedition.org/histoire-cnrs/2661 This "supposed solar furnace", as you term it, finished construction in 1969 and receives 75,000 visitors per year.
> Presumably one of the breakthroughs is that Heliogen can achieve 1000 celsius[sic] at an industrial scale which is not something that appears to have been done before with purely solar energy.
The question of "scale" is essentially one of power — the press release explains that you can expect about one megawatt per acre (250 MW/km² in modern units), and contemplates that you might want to build plants that scale up to two megawatts. To me, that sounds pretty small for industrial scale, but the Odeillo solar furnace is already one megawatt.
Why power rather than volume? Well, you can heat an arbitrarily large or small thing to an arbitrarily high temperature with an arbitrarily large or small amount of power if it's well enough insulated, and modern insulation is very good. But if your heating doesn't have enough power, it will be very slow on a large amount of material, and if it's fighting things like endothermic chemical reactions, it may lose.
Heliogen's press release explains that their pilot plant has 400 mirrors on two acres, so it's probably about two megawatts — twice the size of the 3500° Odeillo, but much lower temperature, and a tiny fraction of the size of existing commercial CSP plants, which run around 600° as explained earlier.
I think you're ignoring a key point of Heliogen's design function. Heliogen is designed to be used directly as the furnace in industrial applications for making steel, cement, etc.
The Odeillo solar furnace can reach higher temperatures but can't transmit them--it can only heat steam, which is a poor transmitter of heat energy to industrial materials.
Ideally, both of these designs would be incorporated together--one to drive the machinery (the Odeillo design) and the other to do the actual physical work of heating the materials (the Heliogen design).
There's nothing to suggest that Heliogen's design is in any way particularly suitable for transmitting heat to somewhere else or for putting a cement kiln or blast furnace at the focus of the mirrors. On the contrary, their pilot plant sites the focus at the top of a tall tower, like many existing CSP power plants, which is a particularly inconvenient place to put a cement kiln, though not unheard of.
For driving machinery you probably want lower temperatures, like the 600° existing coal, nuclear, and CSP power plants use for driving machinery. Coal and nuclear plants could easily generate higher temperatures, but there are a variety of practical difficulties that arise.
Somebody please tell me that I should not lose sleep about this scenario: Take this array of mirrors and mount them to a cloud of drones. Launch it on a sunny day, fly it to the correct position and focus a greater than 1000 degree Celsius beam of photons on your enemies.
The scary thing about this is that it requires nothing more than mirrors, drones and software, and not a deep military industrial complex.
I suppose I should keep this a secret, but experience has shown that no one will take me seriously, so wtf? Why not?
Congratulations, you have just discovered my Evil Plan to Rule the World.
Mwooooha-hah-hah-hah-hah-haaaaaa (evil laugh, eh)
But I'm serious, sort of.
I have a plan to create a huge swarm of flying robots. (Image search Alexander Bell's cellular kites.) They are designed to connect together to make bigger flying structures. There's no upper limit. I will be able to make flying buildings, or what I call the World-Kite...
Anyhow, the thought occurred that I could make the sails reflective and create large (multi-kilometer) solar furnaces. Further, as the swarm will be distributed globally, there's no problem about clouds or night-side: you can just reflect light from the illuminated areas to the desired focal point.
If I manage to do this before anyone catches on I don't see how it could be stopped. I've spent the last N years researching secure computing (I'm not about to let some other hacker usurp my botnet just as I achieve Kardashev Type I status, eh?) and I'm just about to start manufacturing them. (Finally! It's taken about ten years longer than I thought it would.)
- - - -
Imagine being at the focal point of the system as the individual machines line up on you: the sky turning all into Sun...
Why not just arm the drones with machine guns, like the Ziyan Blowfish A3[1], coming soon to arms dealers world wide? The guns will work at night as well!
This, coupled with new software-defined microinverters that work whether the grid is up or down[1], is changing the game worldwide for safe solar AC. I for one am very excited.
This ain't new tech, buzzword + solar tower = headlines. I really love this method of solar though, minimal carbon footprint, ~0% efficiency decay. Coupled w a Sterling engine could even be used for small, off grid energy source.
Will explain why in a later comment, you have to think through the embodied energy flow and materials required to transport etc...as well as solar capacity factor domestically and globally.
Solar panels and Steel Windturbines are manufactured from hydrocarbon feedstock.
Solar and wind drawdowns are powered by natural gas peaking.
Transporting energy intensive Cement and other feedstocks that are manufactured in areas of "renewables viability" are moved with non-renewables sources. (low sulfur bunker fuel maritime ships, diesel-lng-cng trucks, and large earth movers)
Most of the world's volumetric cement consumption happens in areas with low quality/intensity solar.
> Previous commercial concentrating solar thermal systems could only reach temperatures of 565 degrees Celsius, the company said. That’s useful for generating power, but can’t meet the needs of industrial processes.
Or we could build more nuclear power plants, which have the duty cycle necessary for industrial applications. It's no wonder that the more conservative politicians seem to love nuclear more: it's a pro-business, pro-industry carbon-free energy source, unlike wind and solar (or the insane amount of batteries to make it work.
While there are many benefits to nuclear power, this quote is at odds with your comment. Nuclear power suffers a low Carnot efficiency (around 30-40%), as it has an even lower operating temperature than the criticised old CSP design. While nuclear reactors can get arbitrarily hot, practical and safety considerations limit current designs to relatively low (and thus inefficient) temperatures.
Nuclear is also less practical for direct industrial use or distributed deployment because of safety concerns, with only a few exceptional cases like military marine propulsion.
There is only 1 solutions for climate change and 1 that can slow it down fast enuff to give us the time to possible get the only solution to work. Onley Nuclear reactors is clean enuff to slow down the the co2 released in to the atmosphere. 4gram co2/kWh/0.05$ it produces sheep enuff electricity to use carbon catcher and convert energy of co2 to something else. 200gram co2 would set the electricity prices at 0.25$/kWh to customers and we are soon out of drinking water in the world's so 50% on co2 and 50% convert salt water to fresh water will set the electricity price at ~0.45$/kWh. That will buy us time to hopefully figure out fusion before 2080. No one knows if its even possible that we can control that kind of energy to produce electricity for civil use. To just fill the electricity demand the world would need 50 new reactors every week untill 2035 to avoid a global energy crisis, and we would still need fusion before 2080 to start the amount of co2 to go back to as it was in the 1980s before 2200. Renewable energy will never even make a visual mark on a energy production of zero co2 graf and less of one, on the negative co2 graf. Renewable will be a good alternative on some very geografical optimal places that's it. Alot of people this day seems to forget that the E in the formula E=mc2 stands for Energy=everything(mass) both are the same and lay the ground for 1916 general relativity theory. Better known as the law of physics or easier, reality... just run the numbers, GRT doesn't care about money, politics or magic... it's simply predict what work and don't with the same certainty as it predict the time and date for the next solar eclipse.
.png){kind=link}
Given that we know the suns position and that of the mirrors, how is computer vision able to better aim the mirrors?
What am I missing?