science

Auto Added by WPeMatico

DNA Script picks up $38.5 million to make DNA production faster and simpler

DNA Script has raised $38.5 million in new financing to commercialize a process that it claims is the first big leap forward in manufacturing genetic material.

The revolution in synthetic biology that’s reshaping industries from medicine to agriculture rests on three, equally important pillars.

They include: analytics — the ability to map the genome and understand the function of different genes; synthesis — the ability to manufacture DNA to achieve certain functions; and gene editing — the CRISPR-based technologies that allow for the addition or subtraction of genetic code.

New technologies have already been introduced to transform the analytics and editing of genomes, but little progress has been made over the past 50 years in the ways in which genetic material is manufactured. That’s exactly the problem that DNA Script is trying to solve.

Traditionally, making DNA involved the use of chemical compounds to synthesize (or write) DNA in chains that were limited to around 200 nucleotide bases. Those synthetic pieces of genetic code are then assembled to make a gene.

DNA Script’s technology holds the promise of making longer chains of nucleotides by mirroring the enzymatic process through which DNA is assembled within cells — with fewer errors and no chemical waste material. The enzymatic process can accelerate commercial applications in healthcare, chemical manufacturing and agriculture.

“Any technology that can make that faster is going to be very valuable,” says Christopher Voigt, a synthetic biologist at the Massachusetts Institute of Technology in Cambridge, told the journal Nature.

DNA Script isn’t the only company in the market that’s looking to make the leap forward in enzymatic DNA production. Nuclear, a startup working with Harvard University’s famed geneticist, George Church, and Ansa Bio, a startup affiliated with Jay Keasling’s Berkeley lab at the University of California, are also moving forward with the technology.

But the Paris-based company has achieved some milestones that would make its technology potentially the first to come to market with a commercially viable approach.

At least, that’s what new investors LSP and Bpifrance, through its Large Venture fund, are hoping. They’re joined by previous investors Illumina Ventures, M. Ventures, Sofinnova Partners, Kurma Partners and Idinvest Partners in backing the company’s latest funding.

The company said the money would be used to accelerate the development of its first products and establish a presence in the United States.

“As we announced earlier this year at the AGBT General Meeting, DNA Script was the first company to enzymatically synthesize a 200mer oligo de novo with an average coupling efficiency that rivals the best organic chemical processes in use today,”  said Thomas Ybert, chief executive and co-founder of DNA Script. “Our technology is now reliable enough for its first commercial applications, which we believe will deliver the promise of same-day results to researchers everywhere, with DNA synthesis that can be completed in just a few hours.”

Powered by WPeMatico

OpenAI Five crushes Dota2 world champs, and soon you can lose to it too

Dota2 is one of the most popular, and complex, online games in the world, but an AI has once again shown itself to supersede human skill. In matches over the weekend, OpenAI’s “Five” system defeated two pro teams soundly, and soon you’ll be able to test your own mettle against — or alongside — the ruthless agent.

In a blog post, OpenAI detailed how its game-playing agent has progressed from its younger self — it seems wrong to say previous version, since it really is the same extensive neural network as many months ago, but with much more training.

The version that played at Dota2’s premiere tournament, The International, gets schooled by the new version 99 percent of the time. And it’s all down to more practice:

In total, the current version of OpenAI Five has consumed 800 petaflop/s-days and experienced about 45,000 years of Dota self-play over 10 realtime months (up from about 10,000 years over 1.5 realtime months as of The International), for an average of 250 years of simulated experience per day.

To the best of our knowledge, this is the first time an RL [reinforcement learning] agent has been trained using such a long-lived training run.

One is tempted to cry foul at a data center-spanning intelligence being allowed to train for 600 human lifespans. But really it’s more of a compliment to human cognition that we can accomplish the same thing with a handful of months or years, while still finding time to eat, sleep, socialize (well, some of us) and so on.

Dota2 is an intense and complex game with some rigid rules but a huge amount of fluidity, and representing it in a way that makes sense to a computer isn’t easy (which likely accounts partly for the volume of training required). Controlling five “heroes” at once on a large map with so much going on at any given time is enough to tax a team of five human brains. But teams work best when they’re acting as a single unit, which is more or less what Five was doing from the start. Rather than five heroes, it was more like five fingers of a hand to the AI.

Interestingly, OpenAI also discovered lately that Five is capable of playing cooperatively with humans as well as in competition. This was far from a sure thing — the whole system might have frozen up or misbehaved if it had a person in there gumming up the gears. But in fact it works pretty well.

You can watch the replays or get the pro commentary on the games if you want to hear exactly how the AI won (I’ve played but I’m far from good. I’m not even bad yet). I understand they had some interesting buy-back tactics and were very aggressive. Or, if you’re feeling masochistic, you can take on the AI yourself in a limited-time event later this week.

We’re launching OpenAI Five Arena, a public experiment where we’ll let anyone play OpenAI Five in both competitive and cooperative modes. We’d known that our 1v1 bot would be exploitable through clever strategies; we don’t know to what extent the same is true of OpenAI Five, but we’re excited to invite the community to help us find out!

Although a match against pros would mean all-out war using traditional tactics, low-stakes matches against curious players might reveal interesting patterns or exploits that the AI’s creators aren’t aware of. Results will be posted publicly, so be ready for that.

You’ll need to sign up ahead of time, though: The system will only be available to play from Thursday night at 6 PM to the very end of Sunday, Pacific time. They need to reserve the requisite amount of computing resources to run the thing, so sign up now if you want to be sure to get a spot.

OpenAI’s team writes that this is the last we’ll hear of this particular iteration of the system; it’s done competing (at least in tournaments) and will be described more thoroughly in a paper soon. They’ll continue to work in the Dota2 environment because it’s interesting, but what exactly the goals, means or limitations will be are yet to be announced.

Powered by WPeMatico

Talk all things robotics and AI with TechCrunch writers

This Thursday, we’ll be hosting our third annual Robotics + AI TechCrunch Sessions event at UC Berkeley’s Zellerbach Hall. The day is packed start-to-finish with intimate discussions on the state of robotics and deep learning with key founders, investors, researchers and technologists.

The event will dig into recent developments in robotics and AI, which startups and companies are driving the market’s growth and how the evolution of these technologies may ultimately play out. In preparation for our event, TechCrunch’s Brian Heater spent time over the last several months visiting some of the top robotics companies in the country. Brian will be on the ground at the event, alongside Lucas Matney, who will also be on the scene. Friday at 11:00 am PT, Brian and Lucas will be sharing with Extra Crunch members (on a conference call) what they saw and what excited them most.

Tune in to find out about what you might have missed and to ask Brian and Lucas anything else robotics, AI or hardware. And want to attend the event in Berkeley this week? It’s not too late to get tickets.

To listen to this and all future conference calls, become a member of Extra Crunch. Learn more and try it for free.

Powered by WPeMatico

MIT’s ‘cyber-agriculture’ optimizes basil flavors

The days when you could simply grow a basil plant from a seed by placing it on your windowsill and watering it regularly are gone — there’s no point now that machine learning-optimized hydroponic “cyber-agriculture” has produced a superior plant with more robust flavors. The future of pesto is here.

This research didn’t come out of a desire to improve sauces, however. It’s a study from MIT’s Media Lab and the University of Texas at Austin aimed at understanding how to both improve and automate farming.

In the study, published today in PLOS ONE, the question being asked was whether a growing environment could find and execute a growing strategy that resulted in a given goal — in this case, basil with stronger flavors.

Such a task is one with numerous variables to modify — soil type, plant characteristics, watering frequency and volume, lighting and so on — and a measurable outcome: concentration of flavor-producing molecules. That means it’s a natural fit for a machine learning model, which from that variety of inputs can make a prediction as to which will produce the best output.

“We’re really interested in building networked tools that can take a plant’s experience, its phenotype, the set of stresses it encounters, and its genetics, and digitize that to allow us to understand the plant-environment interaction,” explained MIT’s Caleb Harper in a news release. The better you understand those interactions, the better you can design the plant’s lifecycle, perhaps increasing yield, improving flavor or reducing waste.

In this case the team limited the machine learning model to analyzing and switching up the type and duration of light experienced by the plants, with the goal of increasing flavor concentration.

A first round of nine plants had light regimens designed by hand based on prior knowledge of what basil generally likes. The plants were harvested and analyzed. Then a simple model was used to make similar but slightly tweaked regimens that took the results of the first round into account. Then a third, more sophisticated model was created from the data and given significantly more leeway in its ability to recommend changes to the environment.

To the researchers’ surprise, the model recommended a highly extreme measure: Keep the plant’s UV lights on 24/7.

Naturally this isn’t how basil grows in the wild, since, as you may know, there are few places where the sun shines all day long and all night strong. And the arctic and antarctic, while fascinating ecosystems, aren’t known for their flavorful herbs and spices.

Nevertheless, the “recipe” of keeping the lights on was followed (it was an experiment, after all), and incredibly, this produced a massive increase in flavor molecules, doubling the amount found in control plants.

“You couldn’t have discovered this any other way,” said co-author John de la Parra. “Unless you’re in Antarctica, there isn’t a 24-hour photoperiod to test in the real world. You had to have artificial circumstances in order to discover that.”

But while a more flavorful basil is a welcome result, it’s not really the point. The team is more happy that the method yielded good data, validating the platform and software they used.

“You can see this paper as the opening shot for many different things that can be applied, and it’s an exhibition of the power of the tools that we’ve built so far,” said de la Parra. “With systems like ours, we can vastly increase the amount of knowledge that can be gained much more quickly.”

If we’re going to feed the world, it’s not going to be done with amber waves of grain, i.e. with traditional farming methods. Vertical, hydroponic, computer-optimized — we’ll need all these advances and more to bring food production into the 21st century.

Powered by WPeMatico

A startup trying to detect endometriosis through ‘smart tampons’ just landed $9 million in Series A funding

There’s no shortage of so-called femtech startups raising money right now, and little wonder why.

Aside from the growing market opportunity — the global fertility services market is expected to reach $31 billion by 2023, says the consultancy Allied Market Research, nearly double where it stood in 2016 — more women are clamoring for information about their reproductive health, and 15-minute doctor visits aren’t doing the trick.

The newest recipient of venture dollars: NextGen Jane, a 4.5-year-old, Oakland, Calif.-based company that’s hoping to use blood wrung from tampons to find early markers of endometriosis and, later, if all goes well, cervical cancer and other disorders.

The company disclosed just today that it has secured $9 million in Series A funding led by Material Impact, a new fund focused on materials technology that we reported on last November. Other participants in the round include Access Industries, Viking Global Investors, Liminal Ventures and numerous notable angels, including PhDs from Harvard Medical School and Stanford University.

Its approach is far more palatable than the option women have long suffered, which is to have a small camera inserted into their pelvic cavity in search of endometrial cells. (Note: Women typically wind up in this position only after enduring bewildering pain that drives them to see their doctors.) The idea with NextGen Jane instead is for a custom-made tampon to be worn for roughly two hours, placed inside a test tube as part of a home kit and sent to a lab for further analysis.

Of course, it needs to work first, and the technology hasn’t been approved by the FDA. In fact, it hasn’t been proven at all.

The funding could, potentially, make the difference. In an interview with Technology Review last month, NextGen CEO and co-founder Ridhi Tariyal said that a clinical trial is designed and ready to go, but that NextGen Jane needed capital to run a trial on roughly 800 women in order to establish the diagnostic efficacy of menstrual blood. With funding, she’d said, it would take the company about two years to collect a meaningful amount of data.

Reproductive specialists seem torn generally on the trend of startups producing home kits aimed at helping women understand their fertility. While they see some value in the information they provide, they also fear that even home kits that have already been proven to accomplish their intended goal, such as measuring anti-Müllerian hormone, or AMH, may confuse women as much as they help them.

NextGen Jane had previously raised $2.3 million in funding. Meanwhile, TechCrunch has been reporting extensively on the broader uptick in femtech investing. You can find a much deeper dive regarding who has raised what lately and why right here.

Powered by WPeMatico

Ocean drone startup merger spawns Sofar, the DJI of the sea

What lies beneath the murky depths? SolarCity co-founder Peter Rive wants to help you and the scientific community find out. He’s just led a $7 million Series A for Sofar Ocean Technologies, a new startup formed from a merger he orchestrated between underwater drone maker OpenROV and sea sensor developer Spoondrift. Together, they’re teaming up their 1080p Trident drone and solar-powered Spotter sensor to let you collect data above and below the surface. They can help you shoot awesome video footage, track waves and weather, spot fishing and diving spots, inspect boats or infrastructure for damage, monitor acquaculture sites or catch smugglers.

Sofar’s Trident drone (left) and Spotter sensor (right)

“Aerial drones give us a different perspective of something we know pretty well. Ocean drones give us a view at something we don’t really know at all,” former Spoondrift and now Sofar CEO Tim Janssen tells me. “The Trident drone was created for field usage by scientists and is now usable by anyone. This is pushing the barrier towards the unknown.”

But while Rive has a soft spot for the ecological potential of DIY ocean exploration, the sea is crowded with competing drones. There are more expensive professional research-focused devices like the Saildrone, DeepTrekker and SeaOtter-2, as well as plenty of consumer-level devices like the $800 Robosea Biki, $1,000 Fathom ONE and $5,000 iBubble. The $1,700 Sofar Trident, which requires a cord to a surface buoy to power its three hours of dive time and two meters per second speed, sits in the middle of the pack, but Sofar co-founder David Lang things Trident can win with simplicity, robustness and durability. The question is whether Sofar can become the DJI of the water, leading the space, or if it will become just another commoditized hardware maker drowning in knock-offs.

From left: Peter Rive (chairman of Sofar), David Lang (co-founder of OpenROV) and Tim Janssen (co-founder and CEO of Sofar)

Spoondrift launched in 2016 and raised $350,000 to build affordable ocean sensors that can produce climate-tracking data. “These buoys (Spotters) are surprisingly easy to deploy, very light and easy to handle, and can be lowered in the water by hand using a line. As a result, you can deploy them in almost any kind of conditions,” says Dr. Aitana Forcén-Vázquez of MetOcean Solutions.

OpenROV (it stands for Remotely Operated Vehicle) started seven years ago and raised $1.3 million in funding from True Ventures and National Geographic, which was also one of its biggest Trident buyers. “Everyone who has a boat should have an underwater drone for hull inspection. Any dock should have its own weather station with wind and weather sensors,” Sofar’s new chairman Rive declares.

Spotter could unlock data about the ocean at scale

Sofar will need to scale to accomplish Rive’s mission to get enough sensors in the sea to give us more data on the progress of climate change and other ecological issues. “We know very little about our oceans since we have so little data, because putting systems in the ocean is extremely expensive. It can cost millions for sensors and for boats,” he tells me. We gave everyone GPS sensors and cameras and got better maps. The ability to put low-cost sensors on citizens’ rooftops unlocked tons of weather forecasting data. That’s more feasible with Spotter, which costs $4,900 compared to $100,000 for some sea sensors.

Sofar hardware owners do not have to share data back to the startup, but Rive says many customers are eager to. They’ve requested better data portability so they can share with fellow researchers. The startup believes it can find ways to monetize that data in the future, which is partly what attracted the funding from Rive and fellow investors True Ventures and David Sacks’ Craft Ventures. The funding will build up that data business and also help Sofar develop safeguards to make sure its Trident drones don’t go where they shouldn’t. That’s obviously important, given London’s Gatwick airport shutdown due to a trespassing drone.

Spotter can relay weather conditions and other climate data to your phone

“The ultimate mission of the company is to connect humanity to the ocean as we’re mostly conservationists at heart,” Rive concludes. “As more commercialization and business opportunities arise, we’ll have to have conversations about whether those are directly benefiting the ocean. It will be important to have our moral compass facing in the right direction to protect the earth.”

Powered by WPeMatico

Climate change kills off clouds over the ocean in new simulation

We all know climate change is affecting weather systems and ecosystems around the world, but exactly how and in what way is still a topic of intense study. New simulations made possible by higher-powered computers suggest that cloud cover over oceans may die off altogether once a certain level of CO2 has been reached, accelerating warming and contributing to a vicious cycle.

A paper published in Nature details the new, far more detailed simulation of cloud formation and the effects of solar radiation thereupon. The researchers, from the California Institute of Technology, explain that previous simulation techniques were not nearly granular enough to resolve effects happening at the scale of meters rather than kilometers.

These global climate models seem particularly bad at predicting the stratocumulus clouds that hover over the ocean — and that’s a big problem, they noted:

As stratocumulus clouds cover 20% of the tropical oceans and critically affect the Earth’s energy balance (they reflect 30–60% of the shortwave radiation incident on them back to space1), problems simulating their climate change response percolate into the global climate response.

A more accurate and precise simulation of clouds was necessary to tell how increasing temperatures and greenhouse gas concentrations might affect them. That’s one thing technology can help with.

Thanks to “advances in high-performance computing and large-eddy simulation (LES) of clouds,” the researchers were able to “faithfully simulate statistically steady states of stratocumulus-topped boundary layers in restricted regions.” A “restricted region” in this case means the 5×5-km area simulated in detail.

The improved simulations showed something nasty: when CO2 concentrations reached about 1,200 parts per million, this caused a sudden collapse of cloud formation as cooling at the tops of the clouds is disrupted by excessive incoming radiation. Result (as you see at top): clouds don’t form as easily, letting more sun in, making the heating problem even worse. The process could contribute as much as 8 or 10 degrees to warming in the subtropics.

Naturally there are caveats: simulations are only simulations, though this one predicted today’s conditions well and seems to accurately reflect the many processes going on inside these cloud systems (and remember — inherent error could be against us rather than for us). And we’re still a ways off from 1,200 PPM; current NOAA measurements put it at 411 — but steadily increasing.

So it would be decades before this took place, though once it did it would be catastrophic and probably irreversible.

On the other hand, major climatic events like volcanoes can temporarily but violently change these measures, as has happened before; the Earth has seen such sudden jumps in temperature and CO2 levels before, and the feedback loop of cloud loss and resulting warming could help explain that. (Quanta has a great write-up with more context and background if you’re interested.)

The researchers call for more investigation into the possibility of stratocumulus instability, filling in the gaps they had to estimate in their model. The more brains (and GPU clusters) on the case, the better idea we’ll have of how climate change will play out in specific weather systems like this one.

Powered by WPeMatico

Ubiquitilink advance means every phone is now a satellite phone

Last month I wrote about Ubiquitilink, which promised, through undisclosed means, it was on the verge of providing a sort of global satellite-based roaming service. But how, I asked? (Wait, they told me.) Turns out our phones are capable of a lot more than we think: they can reach satellites acting as cell towers in orbit just fine, and the company just proved it.

Utilizing a constellation of satellites in low Earth orbit, Ubiquitilink claimed during a briefing at Mobile World Congress in Barcelona that pretty much any phone from the last decade should be able to text and do other low-bandwidth tasks from anywhere, even in the middle of the ocean or deep in the Himalayas. Literally (though eventually) anywhere and any time.

Surely not, I hear you saying. My phone, that can barely get a signal on some blocks of my neighborhood, or in that one corner of the living room, can’t possibly send and receive data from space… can it?

“That’s the great thing — everybody’s instinct indicates that’s the case,” said Ubiquitilink founder Charles Miller. “But if you look at the fundamentals of the RF [radio frequency] link, it’s easier than you think.”

The issue, he explained, isn’t really that the phone lacks power. The limits of reception and wireless networks are defined much more by architecture and geology than plain physics. When an RF transmitter, even a small one, has a clear shot straight up, it can travel very far indeed.

Space towers

It’s not quite as easy as that, however; there are changes that need to be made, just not anything complex or expensive like special satellite antennas or base stations. If you know that modifying the phone is a non-starter, you have to work with the hardware you’ve got. But everything else can be shaped accordingly, Miller said — three things in particular.

  1. Lower the orbit. There are limits to what’s practical as far as the distance involved and the complications it brings. The orbit needs to be under 500 kilometers, or about 310 miles. That’s definitely low — geosynchronous is 10 times higher — but it’s not crazy either. Some of SpaceX’s Starlink communications satellites are aiming for a similar orbit.
  2. Narrow the beam. The low orbit and other limitations mean that a given satellite can only cover a small area at a time. This isn’t just blasting out data like a GPS satellite, or communicating with a specialized ground system like a dish that can reorient itself. So on the ground you’ll be looking at a 45 degree arc, meaning you can use a satellite that’s within a 45-degree-wide cone above you.
  3. Lengthen the wavelength. Here simple physics come into play: generally, the shorter the wavelength, the less transparent the atmosphere is to it. So you want to use bands on the long (lower Hz) side of the radio spectrum to make sure you maximize propagation.

Having adjusted for these things, an ordinary phone can contact and trade information with a satellite with its standard wireless chip and power budget. But there’s one more obstacle, one Ubiquitilink spent a great deal of time figuring out.

Although a phone and satellite can reach one another reliably, a delay and Doppler shift in the signal due to the speeds and distances involved are inescapable. Turns out the software that runs towers and wireless chips isn’t suited for this; the timings built into the code assume the distance will be less than 30 km, since the curvature of the Earth generally prevents transmitting farther than that.

So Ubiquitilink modified the standard wireless stacks to account for this, something Miller said no one else had done.

“After my guys came back and told me they’d done this, I said, ‘well let’s go validate it,’ ” he told me. “We went to NASA and JPL and asked what they thought. Everybody’s gut reaction was ‘well, this won’t work,’ but then afterwards they just said ‘well, it works.’ ”

The theory became a reality earlier this year after Ubiquitilink launched their prototype satellites. They successfully made a two-way 2G connection between an ordinary ground device and the satellite, proving that the signal not only gets there and back, but that its Doppler and delay distortions can be rectified on the fly.

“Our first tests demonstrated that we offset the Doppler shift and time delay. Everything else is leveraging commercial software,” Miller said, though he quickly added: “To be clear, there’s plenty more work to be done, but it isn’t anything that’s new technology. It’s good solid hardcore engineering, building nanosats and that sort of thing.”

Since his previous company was Nanoracks and he’s been in the business for decades, he’s qualified to be confident on this part. It’ll be a lot of work and a lot of money, but they should be launching their first real satellites this summer. (And it’s all patented, he noted.)

Global roaming

The way the business will work is remarkably simple given the complexity of the product. Because the satellites operate on modified but mostly ordinary off-the-shelf software and connect to phones with no modifications necessary, Ubiquitilink will essentially work as a worldwide roaming operator that mobile networks will pay to access. (Disclosure: Verizon, obviously a mobile network, owns TechCrunch, and for all I know will use this tech eventually. It’s not involved with any editorial decisions.)

Normally, if you’re a subscriber of network X, and you’re visiting a country where X has no coverage, X will have an agreement with network Y, which connects you for a fee. There are hundreds of these deals in play at any given time, and Ubiquitilink would just be one more — except its coverage will eventually be global. Maybe you can’t reach X or Y; you’ll always be able to reach U.

The speeds and services available will depend on what mobile networks want. Not everyone wants or needs the same thing, of course, and a 3G fallback might be practical where an LTE connection is less so. But the common denominator will be data enough to send and receive text at the least.

It’s worth noting also that this connection will be in some crucial ways indistinguishable from other connections: it won’t affect encryption, for instance.

This will of course necessitate at least a thousand satellites, by Miller’s count. But in the meantime, limited service will also be available in the form of timed passes — you’ll have no signal for 55 minutes, then signal for five, during which you can send and receive what may be a critical text or location. This is envisioned as a specialty service at first, then as more satellites join the constellation, that window expands until it’s 24/7 and across the whole face of the planet, and it becomes a normal consumer good.

Emergency fallback

While your network provider will probably charge you the usual arm and leg for global roaming on demand (it’s their prerogative), there are some services Ubiquitilink will provide for free; the value of a global communication system is not lost on Miller.

“Nobody should ever die because the phone in their pocket doesn’t have signal,” he said. “If you break down in the middle of Death Valley you should be able to text 911. Our vision is this is a universal service for emergency responders and global E-911 texting. We’re not going to charge for that.”

An emergency broadcast system when networks are down is also being planned — power outages following disasters are times when people are likely to panic or be struck by a follow-up disaster like a tsunami or flooding, and reliable communications at those times could save thousands and vastly improve recovery efforts.

“We don’t want to make money off saving people’s lives, that’s just a benefit of implementing this system, and the way it should be,” Miller said.

It’s a whole lot of promises, but the team and the tech seem capable of backing them up. Initial testing is complete and birds are in the air — now it’s a matter of launching the next thousand or so.

Powered by WPeMatico

These hyper-efficient solar panels could actually live on your roof soon

The clean energy boffins in their labs are always upping the theoretical limit on how much power you can get out of sunshine, but us plebes actually installing solar cells are stuck with years-old tech that’s not half as good as what they’re seeing. This new design from Insolight could be the one that changes all that.

Insolight is a spinoff from the École Polytechnique Fédérale de Lausanne, where they’ve been working on this new approach for a few years — and it’s almost ready to hit your roof.

Usually solar cells collect sunlight on their entire surface, converting it to electricity at perhaps 15-19 percent efficiency — meaning about 85 percent of the energy is lost in the process. There are more efficient cells out there, but they’re generally expensive and special-purpose, or use some exotic material.

One place people tend to spare no expense, however, is in space. Solar cells on many satellites are more efficient but, predictably, not cheap. But that’s not a problem if you only use just a tiny amount of them and concentrate the sunlight on those; that’s the Insolight insight.

Small but very high-efficiency cells are laid down on a grid, and above that is placed a honeycomb-like lens array that takes light and bends it into a narrow beam concentrated only on the tiny cells. As the sun moves, the cell layer moves ever so slightly, keeping the beams on target. They’ve achieved as high as 37 percent efficiency in tests, and 30 percent in consumer-oriented designs. That means half again or twice the power from the same area as ordinary panels.

Certainly this adds a layer or two of complexity to the current mass-manufactured arrays that are “good enough” but far from state of the art. But the resulting panels aren’t much different in size or shape, and don’t require special placement or hardware, such as a concentrator or special platform. And a recently completed pilot test on an EPFL roof was passed with flying colors.

“Our panels were hooked up to the grid and monitored continually. They kept working without a hitch through heat waves, storms and winter weather,” said Mathiu Ackermann, the company’s CTO, in an EPFL news release. “This hybrid approach is particularly effective when it’s cloudy and the sunlight is less concentrated, since it can keep generating power even under diffuse light rays.”

The company is now in talks with solar panel manufacturers, whom they are no doubt trying to convince that it’s not that hard to integrate this tech with their existing manufacturing lines — “a few additional steps during the assembly stage,” said Ackermann. Expect Insolight panels to hit the market in 2022 — yeah, it’s still a ways off, but maybe by then we’ll all have electric cars too and this will seem like an even better deal.

Powered by WPeMatico

Let’s save the bees with machine learning

Machine learning and all its related forms of “AI” are being used to work on just about every problem under the sun, but even so, stemming the alarming decline of the bee population still seems out of left field. In fact it’s a great application for the technology and may help both bees and beekeepers keep hives healthy.

The latest threat to our precious honeybees is the Varroa mite, a parasite that infests hives and sucks the blood from both bees and their young. While it rarely kills a bee outright, it can weaken it and cause young to be born similarly weak or deformed. Over time this can lead to colony collapse.

The worst part is that unless you’re looking closely, you might not even see the mites — being mites, they’re tiny: a millimeter or so across. So infestations often go on for some time without being discovered.

Beekeepers, caring folk at heart obviously, want to avoid this. But the solution has been to put a flat surface beneath a hive and pull it out every few days, inspecting all the waste, dirt and other hive junk for the tiny bodies of the mites. It’s painstaking and time-consuming work, and of course if you miss a few, you might think the infestation is getting better instead of worse.

Machine learning to the rescue!

As I’ve had occasion to mention about a billion times before this, one of the things machine learning models are really good at is sorting through noisy data, like a surface covered in random tiny shapes, and finding targets, like the shape of a dead Varroa mite.

Students at the École Polytechnique Fédérale de Lausanne in Switzerland created an image recognition agent called ApiZoom trained on images of mites that can sort through a photo and identify any visible mite bodies in seconds. All the beekeeper needs to do is take a regular smartphone photo and upload it to the EPFL system.

The project started back in 2017, and since then the model has been trained with tens of thousands of images and achieved a success rate of detection of about 90 percent, which the project’s Alain Bugnon told me is about at parity with humans. The plan now is to distribute the app as widely as possible.

“We envisage two phases: a web solution, then a smartphone solution. These two solutions allow to estimate the rate of infestation of a hive, but if the application is used on a large scale, of a region,” Bugnon said. “By collecting automatic and comprehensive data, it is not impossible to make new findings about a region or atypical practices of a beekeeper, and also possible mutations of the Varroa mites.”

That kind of systematic data collection would be a major help for coordinating infestation response at a national level. ApiZoom is being spun out as a separate company by Bugnon; hopefully this will help get the software to beekeepers as soon as possible. The bees will thank them later.

Powered by WPeMatico