2026-03-04 08:35:15
The technology, which uses genetically engineered T cells, could target nearly two dozen different solid cancers with one treatment.
Few cancer treatments are as ferocious as CAR T cell therapy.
Often derived from a patient’s own immune cells, CAR T cells are genetically modified to hunt down and destroy cancer cells. The FDA has approved treatments for deadly blood cancers, and treatments tackling autoimmune diseases and preventing tissue scarring in the heart and kidneys have shown promise.
Yet CAR T has struggled against solid tumors. Over 85 percent of cancers fall into this category. Solid tumors have an arsenal of sneaky tactics to evade or deactivate CAR T cells, eventually undermining the treatment.
This month, a Columbia University team broke through one of the barriers with an upgraded design. They engineered a new, ultra-sensitive protein “hook” that seeks out CD70, a protein that dots the surfaces of multiple types of solid cancer cells—but at vastly different levels.
“Some molecules have been identified that are found in 25%, 50%, or 75% of tumor cells,” said study author Michel Sadelain in a press release. “Though a therapy directed at those targets might be successful…you can’t cure somebody if you just eliminate a small fraction or even 90% of their tumor.”
In tests, the supercharged cancer-killers, dubbed HIT cells, detected and wiped out cancer cells with extremely low levels of CD70—so low that the protein was undetectable using traditional methods. In kidney, ovarian, and pancreatic cancer grown from patients’ cells in petri dishes and in mouse models, HIT completely eliminated all signs of these tumors.
Like CAR T, the new approach is plug-and-play. The protein hook can be redesigned to target other faint cancer protein markers that have previously escaped detection.
“We hope our CD70-directed HIT cells help us find a way to eradicate the entire tumor,”
said study author Sophie Hanina.
Our immune system naturally fights off cancer. T cells, for example, roam the body looking for threats. When they identify cancerous cells, they signal other immune cells to launch a coordinated effort to wipe out the cancer before it expands.
The identification process relies on antigens, proteins that dot the surfaces of cancer cells like beacons. But tumors are highly versatile and rapidly evolve their antigen signature, essentially cloaking themselves from immune attacks.
CAR T cells override the defense. Here, T cells are extracted from a patient’s body and genetically engineered with custom-designed protein hooks to grab onto cancer antigens.
Multiple blood cancers have a heavy coat of a single shared protein on their surfaces, making them a perfect target for CAR T therapy. Solid tumors, however, are different. Tumors are dotted with a wide range of antigens, many of which are present in normal tissues. This increases the chances CAR T might attack healthy cells and reduces its effectiveness.
Even for the same antigen, some cells in solid tumors express high levels, others very low. The latter escape CAR T detection and linger as a reservoir that can regrow the tumor.
For a persistent solid cancer cure, “you have to get down to the very last cell,” said Sadelain.
An ideal target antigen needs to check two boxes: It’s expressed across multiple tumor cell types, and at the same time, it’s absent in normal cells.
The antigen in the new study, CD70, fits the bill. It occurs in a variety of solid cancers, making it a valuable target beacon. But previous attempts targeting CD70 struggled to control cancer in clinical trials. This is partly because cancer cells within a single tumor have different levels of the antigen, and some seemingly lack the marker altogether, allowing them to escape detection.
But are these cancer cells truly devoid of the antigen, or is it just that scientists, and the CAR T cells they’ve engineered, can’t find them using current methods?
Researchers can see most proteins under the microscope but only if they’re at high enough levels. Rather than relying on conventional imaging, the team looked for CD70 gene expression in donated cancer patient samples. These lab models mimic the complexity of solid tumors.
CD70 antigens dotted each cell in multiple tumors, although at different levels of intensity. “We found that apparent CD70-negative tumor cells do in fact express low levels of CD70, though not at a level high enough to be eliminated by conventional CAR T cells,” wrote the team.
Taking aim at cancer cells with faint CD70 levels, the team tapped into their previous work genetically engineering cells to detect low-level antigens. The hooks on these HIT cells mimic those from a population of highly sensitive T cells naturally found in our bodies.
The team redesigned HIT cells to specifically target CD70. Because normal cells don’t use this molecular pathway, HIT cells largely ignored them, lowering the risk of collateral damage.
“HIT cells are the next generation of CAR T cells. They can be programmed like a CAR T cell, but they have the sensitivity of a natural T cell and can detect cancer cells that have only a vanishingly small number of target molecules,” said Hanina.
Ovarian and pancreatic cancer cells have mixed levels of CD70. Several tests in highly aggressive models for these cancers found that HIT cells completely eradicated the tumors in petri dishes. The treatment also cleared cancer cells in different types of solid tumors in mice, even ones with low CD70 levels. Conventional CAR T only eliminated a fraction of the cancer.
A recent CAR T clinical trial targeting CD70 found CAR T cells could infiltrate and linger near kidney tumors, but their effectiveness was based on detection, which varied depending on the number of CD70 beacons. Because HIT cells are more sensitive, they could hunt down and kneecap more cancer cells.
But HIT cells may have side effects. Although CD70 isn’t expressed in most healthy tissues, its level skyrockets in immune cells during infections, which could trigger friendly fire. The team plans to investigate the treatment’s safety and efficacy in patients with ovarian cancer at the Columbia University Irving Medical Center.
If successful, the technology could benefit roughly 20 other types of solid cancer that express CD70, including deadly brain cancers such as glioblastoma.
“Curing solid tumors is not easy, but this work solves one piece of the puzzle,” said Sadelain.
The post These Supercharged Immune Cells Completely Eliminated Solid Tumors in Mice appeared first on SingularityHub.
2026-03-03 07:47:26
The flower-shaped device wraps around brain organoids and maps nearly all electrical activity.
Pea-sized brain blobs are a chatty bunch. Packed with neurons that spark with electrical activity, brain organoids—or “mini brains”—are a now popular way to study the human brain.
Some organoids model the brain’s wiring during early development. Others, made from patients’ skin cells, retain DNA mutations that could lead to schizophrenia or autism. Scientists studying mini brains hope to find patterns associated with diseases and test ways to fix them.
But it’s harder to record activity in organoids than neurons lying flat in a dish. Traditional electrode arrays are too large and stiff. More recent soft electronics are a better fit, but they can only record a fraction of each organoid at a time. This makes it difficult to figure out how all the neurons work together and could miss key aspects of the way the brain functions as a whole.
Recently, a team at Northwestern University and collaborators offered up a unique solution: A pop-up electrode mesh that envelopes entire mini brains. Each device starts out as a flower-like lattice before transforming into a breathable 3D net that gently wraps around brain organoids. The device’s 240 microelectrodes simultaneously capture electrical activity from nearly the entirety of an organoid’s surface, providing a birds-eye view of overall function.
Using the device, the team detected brain-wave-like electrical oscillations rippling across the organoids. Thanks to its porous design, the mesh allows nutrients—and drugs—to flow through. In several tests using drugs known to spike or lower neural activity, the device readily picked up changes across the mini brain, hinting at its potential as a drug testing platform.
“This advance is really about building the right tools…we can now record from and stimulate hundreds of locations across [an organoid’s] surface at once. This allows us to study neural activity at the level of whole networks rather than isolated signals,” said study author Colin Franz in a press release.
Technologies that tap into and alter brain activity have exploded in variety and efficiency. Some sit on top of the brain, under the skull, to monitor large areas. Others directly record a vast number of single neurons as they fire away. Trusty non-invasive methods like electroencephalograms (EEG) have mapped whole-brain activity for over a century.
Recording from mini brains is different. Recording devices meant for the human brain are far too bulky. Those designed for cells in petri dishes break if they’re bent.
“Integrated circuits in consumer electronics are perfectly planar, sitting on wafer-based substrates,” said study author John Rogers. “That conventional layout represents a very significant geometrical mismatch relative to the spherical shapes of these organoids.”
Scientists have recently turned to mesh-like electrode systems that are more flexible, such as a basket-like design inspired by Japanese paper folding. Another, shaped like a flower, turns into an electrode-studded claw to grasp organoids. Most of these can record single cells as the organoids develop without changing their shape or cellular and genetic makeup.
Researchers already use these designs to eavesdrop on a variety organoids. But the devices struggle to capture whole-brain dynamics. Some have just a few dozen electrodes; others cover only a small region. To truly decipher mini brain activity, scientists need full-coverage hardware.
The new device is like a draw-string coin purse. It starts out completely flat like a flower and then gradually cinches itself into a soft, flexible mesh that envelops the mini brain.
Matching the mesh to the organoid’s shape without damage is one challenge. Another is to ensure all 240 electrodes are distributed across a mini brain’s entire surface.
The team turned to a design that allows electrodes to move in predictable ways as the device changes shape. When flat, the device is smaller than a US quarter. Once converted into a 3D purse, its electrodes are evenly spread out. Each one is only 10 microns across, roughly the size of a single cell. The shape-shifting mechanism works like a pop-up book. Each petal of the flower is carefully tailored to bend into differently sized purses and shapes to minimize gaps between the electrodes and the organoid surface.
The device’s mesh allows nutrients to flow into the organoid but prevents cells from spreading past the pores and beyond the reach of the electrodes.
“The device’s structure needs to support these metabolic processes to sustain the viability of the tissue,” said Rogers. “Basically, the organoid needs to breathe. The hardware must not significantly constrain or suffocate it.”
The device covered roughly 91 percent of the surface of an average-sized mini brain after 60 days in culture. Because the team knew each electrode’s location, they were able to reconstruct the neural oscillations to create a 3D widescreen view of the organoid’s activity.
In one case, triggering activity in a small region led to highly synchronized, wave-like activity across most surface neurons. The activity patterns, previously undetectable in brain organoids, mimicked those seen in developing human brains.
These synchronized waves broke down dramatically after a dose of botulinum toxin, which is known to dampen brain activity by inhibiting chemical connections between neurons. This suggests their wiring is similar to that of our brains.
The device also captured brain-wide activity changes due to a neurochemical imbalance. Watching these kinds of changes could help scientists study a wide range of diseases such as Parkinson’s, multiple sclerosis, and amyotrophic lateral sclerosis (ALS).
The system also worked with other established observation methods. For example, it reliably recorded electrical signals in tandem with a technology that tracks brain chemicals. The system also responded to light beams activating groups of neurons (an approach called optogenetics).
Mixing and matching different strategies, including wholly new approaches, like the one in this study, could yield a more in-depth understanding of mini brains—and in turn, our own.
The post New Device Detects Brain Waves in Mini Brains Mimicking Early Human Development appeared first on SingularityHub.
2026-02-28 23:00:00
Breaking Encryption With a Quantum Computer Just Got 10 Times EasierKarmela Padavic-Callaghan | New Scientist ($)
“In 2019, Craig Gidney at Google Quantum AI co-authored a paper that reduced [the requirement to break RSA encryption] from 170 million to 20 million quantum bits, or qubits. And in 2025, Gidney devised a way to slash that number to less than a million qubits. Now, Paul Webster at Iceberg Quantum in Australia and his colleagues have managed to decrease the number even further to about 100,000 qubits.”
Tech Has Never Caused a Job Apocalypse. Don’t Bet on It Now.Greg Ip | The Wall Street Journal ($)
“No one should dismiss any scenario, even the most dystopian, with high conviction. Certainly not journalists, whose way of life is in AI’s crosshairs. But I keep stumbling over one small problem with the doomsday vision: It requires a breakdown in how the market economy functions. Nothing like it has happened in the US before, and there is no evidence it is happening now.”
A Recent 3D Printing Breakthrough Brings Us One Step Closer to You Downloading a CarJustin Caffier | Gizmodo
“A team at [MIT] has recently developed a printer with four different extruders that outputs five different materials to produce a fully functioning linear motor in about three hours. …The team explained how by retrofitting a printer with enough extruders to handle the various materials needed to make a working motor, they decimated the usual production time for such a device and brought the material costs down to around $0.50.”
Human Brain Cells on a Chip Learned to Play Doom in a WeekAlex Wilkins | New Scientist ($)
“A clump of human brain cells can play the classic computer game Doom. While its performance is not up to par with humans, experts say it brings biological computers a step closer to useful real-world applications, like controlling robot arms.”
Waymo Robotaxis Are Now Operating in 10 US CitiesKirsten Korosec | TechCrunch
“‘Waymo is serving more riders than ever, as we are on track to serve over one million rides per week by the end of this year,’ Mawakana said in a blog post Tuesday, adding that the company is laying the groundwork for robotaxi service in more than 20 cities.”
AI Will Never Be ConsciousMichael Pollan | Wired ($)
“By the time I finished digesting the Butlin report, the Copernican moment I’d worried about seemed more distant than the report’s bold conclusion had led me to believe. After reviewing the half‑dozen or so theories of consciousness covered by the report, it seemed clear that all of them stacked the deck by taking for granted that consciousness could be reduced to some kind of algorithm.”
Andrew Ng Says AGI Is Decades Away—and the Real AI Bubble Risk Is in the Training LayerVictor Dey | Fast Company ($)
“Maybe a year ago, AGI felt 50 years away. Over the past year, perhaps we’ve made a solid 2% of progress, with another 49 years to go. These numbers are metaphorical, so don’t take them too seriously. [Laughs] But we are closer than before, yet many decades away from an AI that matches human intelligence. If you stick with the original definition—aligned with what people genuinely imagine AGI to be—we remain very, very far away.”
The Hidden Cost of Letting AI Make Your Life EasierShai Tubali | Big Think
“‘[AI] is designed to take over tasks that are effortful for us,’ Nyholm says. …The difficulty, he adds, is that many effortful tasks are precisely the ones that carry meaning. Deep relationships require patience, friction, and vulnerability. Skills demand time, frustration, and persistence. …When effort, creativity, and skill fall away, meaningfulness no longer seems the right category.”
The Human Work Behind Humanoid Robots Is Being HiddenJames O’Donnell | MIT Technology Review ($)
“The roboticist Aaron Prather told me about recent work with a delivery company that had its workers wear movement-tracking sensors as they moved boxes; the data collected will be used to train robots. The effort to build humanoids will likely require manual laborers to act as data collectors at massive scale. ‘It’s going to be weird,’ Prather says. ‘No doubts about it.'”
The 5 Biggest Obstacles to AI Data Centers in SpaceEthan Siegel | Big Think
“Is this an example of an emerging technology that could provide an off-world solution to the problem of competing demands for limited resources? Or is it, like the hyperloop, an example of grift: where the concept itself isn’t exactly physically impossible, but is rendered so impractical due to the actual physical constraints of the endeavor, that it absolutely cannot materialize as advertised?”
Meta Director of AI Safety Allows AI Agent to Accidentally Delete Her InboxMatthew Gault | 404 Media
“As countless people on X have said in response to her post, seeing the person in charge of making sure powerful AI tools are safe at one of the biggest tech companies in the world trust an AI agent that is known to pose several serious security risks, does not inspire a lot of confidence in what Meta and other big AI companies are doing.”
The post This Week’s Awesome Tech Stories From Around the Web (Through February 28) appeared first on SingularityHub.
2026-02-27 23:00:00
We need to give models knowledge that anchors their behavior to the realities of our world.
Modern AI chatbots can do amazing things, from writing research papers to composing Shakespearian sonnets about your cat. But amid the sparks of genius, there are flashes of idiocy. Time and again, the large language models, or LLMs, behind today’s generative AI tools make basic errors—from failing to solve basic high school math problems to stumbling over the rules of Connect Four.
This instability has been called “jagged intelligence” in tech circles, and it isn’t just a quirk—it’s a critical failing and part of the reason many experts believe we’re in an AI bubble. You wouldn’t hire a doctor or lawyer who, despite giving sound medical or legal advice, sometimes acts like they are clueless about how the world works. Enterprises seem to feel the same way about putting “jagged” AI in charge of supply chains, HR processes, or financial operations.
To solve the jagged intelligence problem, we must give our AI models access to a more powerful, more structured, and ultimately far more human stock of knowledge. Having engineered a range of AI systems over 30 years, I have found such knowledge to be an indispensable component of any reliable system.
This is because the technological innovations that launched the AI era aren’t capable of smoothing out these jagged edges. Current AI models don’t possess clear rules about how the world works; instead, they infer things from vast pools of data. In other words, they don’t know things, so they’re forced to guess—and when they guess wrong, the results range from the comical to the catastrophic.
Think about how humans learn. Born into “blooming, buzzing confusion,” babies spot patterns in the world around them: Faces are fun to look at, mom smells great, the cat scratches if you yank its tail. But pattern recognition is soon supplemented by clearly articulated knowledge: rules we’re taught, rather than things we absorb. From ABCs to arithmetic to how to load a dishwasher or drive a car, we use codified knowledge to learn efficiently—and avoid idiotic or dangerous mistakes along the way.
Current AI models don’t possess clear rules about how the world works; instead, they infer things from vast pools of data.
Frontier AI labs are already dabbling in this approach. Early LLMs struggled with grade-school math, so researchers bolted on actual mathematical knowledge—not hazy inferences, but explicit rules about how math works. The result: Google’s latest models can now reliably solve math Olympiad problems.
Adding more data of different types—for example video data, being advocated by AI luminaries such as Yann LeCun—won’t overcome the fundamental challenge of jagged intelligence. Even with extra data, it’s mathematically certain that the models will keep making mistakes—because that’s how probabilistic, data-driven AI works. Instead, we need to give models knowledge—rigidly described concepts and constraints, rules and relationships—that anchor their behavior to the realities of our world.
To give AI models a human stock of knowledge, we need to rapidly build a public database of formal knowledge spanning a range of disciplines. Of course, the rules of math are clear; the workings of other fields—health care, law, economics, or education, say—are, in some ways, vastly more complex. This challenge is now within our reach, as the growth of companies such as Scale AI, which provides high-quality data for training AI models, points to the emergence of a new profession—one that translates human expertise into machine-readable form and, in doing so, shapes not just what AI can do, but what it comes to treat as true.
This knowledge base could be accessed on demand by developers (or even AI agents) to provide verifiable insights covering everything from loading a dishwasher to the intricacies of the tax code. AI models would make fewer absurd mistakes, because they wouldn’t need to deduce everything from first principles. (Some research also suggests that such models would require far less data and energy, though these claims have yet to be proven.)
Unlike today’s opaque AI models, whose knowledge emerges from pattern recognition and is spread across billions of parameters, a formally distilled body of human knowledge could be directly examined, understood, and controlled. Regulators could verify a model’s knowledge, and users could ensure that tools were mathematically guaranteed not to make idiotic mistakes.
We need to give models knowledge—rigidly described concepts and constraints, rules and relationships—that anchors their behavior to the realities of our world.
The ambition to create such a knowledge resource is nothing new in AI. Even though previous efforts produced inconclusive results, it’s time to make a fresh start. Much as biologists use algorithms to speedrun the once-laborious process of modeling proteins, AI researchers could leverage generative AI to aid knowledge modeling.
It’s clear that current AI models are getting smarter and will get better by using different data. And yet, to overcome the challenge of jagged intelligence—and turn AI models into trusted partners and true drivers of value—we need to redefine the way models relate to and learn about the world. Data-driven algorithms allowed us to start talking to machines. But knowledge, not data, is the key to sustaining the future of AI past the potential bubble.
This article was originally published on Undark. Read the original article.
The post Sparks of Genius to Flashes of Idiocy: How to Solve AI’s ‘Jagged Intelligence’ Problem appeared first on SingularityHub.
2026-02-27 07:28:40
Fossil fuels still dominate the energy mix. But growth in renewables offset nearly 75% of new power demand.
Booming energy demand is driving a scramble to set up new generating capacity, and one technology is proving to be the clear winner. Newly released federal data shows that solar power grew by more than 35 percent year-over-year in 2025, outpacing all other forms of generation.
After decades of relatively flat electricity use, US power demand is rising again. A new report from the US Energy Information Administration shows that consumption jumped 2.8 percent in 2025, thanks to rising industrial activity and the rapid expansion of energy-hungry AI data centers.
While increased fossil-fuel generation met much of that additional demand—a revival in coal, in particular—solar power posted the fastest growth of any major source. According to an analysis by Ars Technica, the Energy Information Administration data shows US solar generation increased by more than 35 percent year-over-year, driven by 27 gigawatts of newly installed capacity.
The surge pushed total solar output above hydroelectric power for the first time in terms of total annual generation. Hydropower output itself was relatively stable compared with the prior year, while solar continued its recent breakneck expansion, with capacity increasing rapidly across multiple regions.
Solar’s surge, which added about 85 terawatt-hours of generation, met about two-thirds of the increased energy demand. This number rose to 73 percent when combined with wind power, which grew by a more modest 2.8 percent.
But the data’s not-so-silver lining is that the remaining demand was met primarily by 13 percent growth in coal power. That bucks the recent trend of coal’s diminishing importance in the US power supply and was driven by a complex confluence of changes in the US energy system.
In previous years, natural gas has been the go-to fossil fuel due to abundant domestic supply and the ability to rapidly ramp power up and down. However, the Trump administration’s tariff policies have made it more difficult and expensive to source the equipment for gas power plants, and rapid expansion in gas exports means domestic utilities are competing with foreign buyers for fuel.
Altogether, this made gas generation a less reliable bet, and generation actually shrank 3.3 percent last year. That made coal a more attractive option, and its position in the US energy mix grew.
Still, the outlook remains bright for renewables, and solar in particular. A recent Energy Information Administration analysis showed roughly 43 gigawatts of utility-scale solar capacity is planned or under construction for 2026, potentially making it an even bigger year for solar than 2025. More than half of this new capacity is in just four states: Texas, Arizona, California, and Michigan.
And wind power’s growth could more than double this year with planned additions of 11.8 gigawatts, 60 percent of which are in New Mexico, Texas, Illinois, and Wyoming.
Growth in renewables will also be supported by another record-breaking year for added battery storage. Last year, the industry added a record 15 gigawatts of capacity to the grid. Planned projects would grow capacity by another 24 gigawatts in 2026, 80 percent of which will be in Texas, California, and Arizona.
It’ll be a while yet before green energy overtakes fossil fuels, which still accounted for 58 percent of total generation in 2025. But the data suggests the energy transition is well underway. If solar and wind continue to meet growing energy demand, the US energy system could soon look very different.
The post US Solar Surged 35% in 2025, Overtaking Hydro for the First Time appeared first on SingularityHub.
2026-02-25 05:50:18
Scientists are co-opting seismic sensors to detect space debris streaking through the atmosphere at hypersonic speeds.
In the early morning of April 2, 2024, the sky over southern California lit up with flashes of blazing light. Residents were bewildered. Were they missiles? A crashing plane? The unusual activity confused even experts—until they realized it was a disposable part of China’s Shenzhou-15 spacecraft burning up in the atmosphere as it returned to Earth.
Scientists knew the event was on the horizon and had mapped out a potential entry point over the northern Atlantic Ocean, thousands of miles from metropolitan Los Angeles. Luckily, no one was hurt as the module broke apart over the city.
But the incident underlined an uncomfortable truth. We’re nowhere near being able to accurately predict the path of space debris as it rains down. As more spacecraft are launched and reenter the atmosphere, damage to infrastructure and Earthlings is only a matter of time.
Researchers are looking into a solution from an unexpected source: sensors that measure earthquakes. As space debris plummets to the ground at hypersonic speeds, it generates a sonic boom. This causes a slight tremor in the ground that the sensors readily register.
Using data from a network of these sensors, Benjamin Fernando at Johns Hopkins University and Constantinos Charalambous at Imperial College London developed a system that can reconstruct the path of space debris with unprecedented accuracy. They used the system to map Shenzhou-15’s speed, altitude, gradual disintegration, and final destination.
To be clear, this isn’t an early warning system. Because sonic booms lag behind the objects causing them, the method is like a forensic reconstruction of space debris’ final journey. Still, it can quickly identify potential fall-out zones for faster retrieval and cleanup, which is especially important if the junk is toxic or radioactive.
The work is “a crucial step toward near-real-time monitoring of natural and anthropogenic objects entering from space,” wrote Chris Carr at the Los Alamos National Laboratory, who was not involved in the work.
Launching satellites was once a colossal undertaking. But thanks to innovations by SpaceX and national space agencies across the world, it’s becoming far more routine.
These spacecraft have already changed life on Earth. Thousands of Starlink satellites beam the internet to previous dead zones and disaster areas. Miniature satellites are now an affordable research platform scientists use to profile weather, measure solar winds, and track the effects of microgravity and radiation on living cells. And a new space race will only grow the fleets of spacecraft already blanketing the Earth.
“The big change that we’ve seen since 2020 is the rise of satellite mega-constellations…companies not putting up a dozen spacecraft, but maybe a thousand or ten thousand over the course of a few years,” Fernando told Science.
Mega-constellations have already caused problems for scientists by polluting astronomical images with bright streaks. They may also increase the rate at which space debris rains down. In a paper describing their system, Fernando and Charalambous write that in 2025 there were roughly four to five re-entries a day, and the numbers are likely to rapidly grow.
We already monitor spacecraft in orbit. Telescopes bring real-time visuals. Radar tracks location and speed. But these tools struggle as a spacecraft drifts into the Earth’s upper atmosphere.
The interaction between fragments and air becomes “really chaotic,” said Fernando. “We can no longer predict with particularly good accuracy exactly where [and when] a piece of re-entering space debris is going to enter the atmosphere.”
Radar can track spacecraft parts as they return to Earth, but the technology is limited to small regions of the world and barely covers the oceans. Even when we know the final fate of a piece of debris, it’s often difficult to reconstruct its full trajectory.
The new work was inspired by the way scientists track meteoroids using a dense network of earthquake sensors to detect tiny vibrations in the ground.
The Shenzhou-15 capsule entered the atmosphere going roughly 25 to 30 times the speed of sound. Like a fighter jet, it triggered a powerful sonic boom roughly 80 kilometers (50 miles) above the ground. The boom traveled to Earth’s surface where seismic sensors detected it.
It’s like picking up an earthquake, only “in this case the waves are coming from up versus with earthquakes they tend to come from down,” said Fernando.
Southern California is heavily dotted with seismic sensors, each measuring activity in a small area. To model the spacecraft’s path and speed, the team compiled the largest sonic boom each sensor registered and its arrival time and compiled the data into a map.
The map captured where, when, and how the capsule broke down as it hurtled through the atmosphere. Earlier on, the sensors recorded large, discrete signals. These later became more scattered and complex, suggesting the capsule gradually disintegrated rather than blowing up all at once.
The results are “consistent with on-ground observations, including videos and witness reports of multiple fireballs flying across the sky,” wrote Carr. After more deeply combing through the data, the team showed it could also be used to measure the size of each piece of decaying debris.
The spacecraft’s sonic signature differed from those generated by meteorites, making it possible to tease apart human-made objects and those of natural origins.
Differentiating the two categories is key. Meteorites pose “kinetic risk” as chunks slam into the ground, damaging cars, houses, and other infrastructure. Human space debris, however, could also contain metals, toxic or flammable material, or in rare cases, radioactive components. The model also reconstructed how different parts of the spacecraft disintegrated, potentially making it easier to predict whether chunks have burned up completely in the atmosphere or have reached the ground, making it useful for recovery or clean-up missions.
Crash-and-burn isn’t a spacecraft’s only destiny. Engineers are also working to move defunct satellites into higher orbits that would be stable for “thousands of years” according to Fernando, though this doesn’t solve the space junk problem. Other researchers are exploring ways to design spacecraft such that they completely burn up both safely and predictably.
For now, the technology works best in places with lots of seismic sensors, which are rare. But there’s a push to add sensors in places that are vulnerable due to sensitive ecology or geology at prices far lower than building radar systems to track re-entry, said Fernando.
The post More Space Junk Is Plummeting to Earth. Earthquake Sensors Can Track It by the Sonic Booms. appeared first on SingularityHub.
