Tag Archives: harvard

#439280 Google and Harvard Unveil the Largest ...

Last Tuesday, teams from Google and Harvard published an intricate map of every cell and connection in a cubic millimeter of the human brain.

The mapped region encompasses the various layers and cell types of the cerebral cortex, a region of brain tissue associated with higher-level cognition, such as thinking, planning, and language. According to Google, it’s the largest brain map at this level of detail to date, and it’s freely available to scientists (and the rest of us) online. (Really. Go here. Take a stroll.)

To make the map, the teams sliced donated tissue into 5,300 sections, each 30 nanometers thick, and imaged them with a scanning electron microscope at a resolution of 4 nanometers. The resulting 225 million images were computationally aligned and stitched back into a 3D digital representation of the region. Machine learning algorithms segmented individual cells and classified synapses, axons, dendrites, cells, and other structures, and humans checked their work. (The team posted a pre-print paper about the map on bioArxiv.)

Last year, Google and the Janelia Research Campus of the Howard Hughes Medical Institute made headlines when they similarly mapped a portion of a fruit fly brain. That map, at the time the largest yet, covered some 25,000 neurons and 20 million synapses. In addition to targeting the human brain, itself of note, the new map includes tens of thousands of neurons and 130 million synapses. It takes up 1.4 petabytes of disk space.

By comparison, over three decades’ worth of satellite images of Earth by NASA’s Landsat program require 1.3 petabytes of storage. Collections of brain images on the smallest scales are like “a world in a grain of sand,” the Allen Institute’s Clay Reid told Nature, quoting William Blake in reference to an earlier map of the mouse brain.

All that, however, is but a millionth of the human brain. Which is to say, a similarly detailed map of the entire thing is yet years away. Still, the work shows how fast the field is moving. A map of this scale and detail would have been unimaginable a few decades ago.

How to Map a Brain
The study of the brain’s cellular circuitry is known as connectomics.

Obtaining the human connectome, or the wiring diagram of a whole brain, is a moonshot akin to the human genome. And like the human genome, at first, it seemed an impossible feat.

The only complete connectomes are for simple creatures: the nematode worm (C. elegans) and the larva of a sea creature called C. intestinalis. There’s a very good reason for that. Until recently, the mapping process was time-consuming and costly.

Researchers mapping C. elegans in the 1980s used a film camera attached to an electron microscope to image slices of the worm, then reconstructed the neurons and synaptic connections by hand, like a maddeningly difficult three-dimensional puzzle. C. elegans has only 302 neurons and roughly 7,000 synapses, but the rough draft of its connectome took 15 years, and a final draft took another 20. Clearly, this approach wouldn’t scale.

What’s changed? In short, automation.

These days the images themselves are, of course, digital. A process known as focused ion beam milling shaves down each slice of tissue a few nanometers at a time. After one layer is vaporized, an electron microscope images the newly exposed layer. The imaged layer is then shaved away by the ion beam and the next one imaged, until all that’s left of the slice of tissue is a nanometer-resolution digital copy. It’s a far cry from the days of Kodachrome.

But maybe the most dramatic improvement is what happens after scientists complete that pile of images.

Instead of assembling them by hand, algorithms take over. Their first job is ordering the imaged slices. Then they do something impossible until the last decade. They line up the images just so, tracing the path of cells and synapses between them and thus building a 3D model. Humans still proofread the results, but they don’t do the hardest bit anymore. (Even the proofreading can be refined. Renowned neuroscientist and connectomics proponent Sebastian Seung, for example, created a game called Eyewire, where thousands of volunteers review structures.)

“It’s truly beautiful to look at,” Harvard’s Jeff Lichtman, whose lab collaborated with Google on the new map, told Nature in 2019. The programs can trace out neurons faster than the team can churn out image data, he said. “We’re not able to keep up with them. That’s a great place to be.”

But Why…?
In a 2010 TED talk, Seung told the audience you are your connectome. Reconstruct the connections and you reconstruct the mind itself: memories, experience, and personality.

But connectomics has not been without controversy over the years.

Not everyone believes mapping the connectome at this level of detail is necessary for a deep understanding of the brain. And, especially in the field’s earlier, more artisanal past, researchers worried the scale of resources required simply wouldn’t yield comparably valuable (or timely) results.

“I don’t need to know the precise details of the wiring of each cell and each synapse in each of those brains,” nueroscientist Anthony Movshon said in 2019. “What I need to know, instead, is the organizational principles that wire them together.” These, Movshon believes, can likely be inferred from observations at lower resolutions.

Also, a static snapshot of the brain’s physical connections doesn’t necessarily explain how those connections are used in practice.

“A connectome is necessary, but not sufficient,” some scientists have said over the years. Indeed, it may be in the combination of brain maps—including functional, higher-level maps that track signals flowing through neural networks in response to stimuli—that the brain’s inner workings will be illuminated in the sharpest detail.

Still, the C. elegans connectome has proven to be a foundational building block for neuroscience over the years. And the growing speed of mapping is beginning to suggest goals that once seemed impractical may actually be within reach in the coming decades.

Are We There Yet?
Seung has said that when he first started out he estimated it’d take a million years for a person to manually trace all the connections in a cubic millimeter of human cortex. The whole brain, he further inferred, would take on the order of a trillion years.

That’s why automation and algorithms have been so crucial to the field.

Janelia’s Gerry Rubin told Stat he and his team have overseen a 1,000-fold increase in mapping speed since they began work on the fruit fly connectome in 2008. The full connectome—the first part of which was completed last year—may arrive in 2022.

Other groups are working on other animals, like octopuses, saying comparing how different forms of intelligence are wired up may prove particularly rich ground for discovery.

The full connectome of a mouse, a project already underway, may follow the fruit fly by the end of the decade. Rubin estimates going from mouse to human would need another million-fold jump in mapping speed. But he points to the trillion-fold increase in DNA sequencing speed since 1973 to show such dramatic technical improvements aren’t unprecedented.

The genome may be an apt comparison in another way too. Even after sequencing the first human genome, it’s taken many years to scale genomics to the point we can more fully realize its potential. Perhaps the same will be true of connectomics.

Even as the technology opens new doors, it may take time to understand and make use of all it has to offer.

“I believe people were impatient about what [connectomes] would provide,” Joshua Vogelstein, cofounder of the Open Connetome Project, told the Verge last year. “The amount of time between a good technology being seeded, and doing actual science using that technology is often approximately 15 years. Now it’s 15 years later and we can start doing science.”

Proponents hope brain maps will yield new insights into how the brain works—from thinking to emotion and memory—and how to better diagnose and treat brain disorders. Others, Google among them no doubt, hope to glean insights that could lead to more efficient computing (the brain is astonishing in this respect) and powerful artificial intelligence.

There’s no telling exactly what scientists will find as, neuron by synapse, they map the inner workings of our minds—but it seems all but certain great discoveries await.

Image Credit: Google / Harvard Continue reading

Posted in Human Robots

#439000 Can AI Stop People From Believing Fake ...

Machine learning algorithms provide a way to detect misinformation based on writing style and how articles are shared.

On topics as varied as climate change and the safety of vaccines, you will find a wave of misinformation all over social media. Trust in conventional news sources may seem lower than ever, but researchers are working on ways to give people more insight on whether they can believe what they read. Researchers have been testing artificial intelligence (AI) tools that could help filter legitimate news. But how trustworthy is AI when it comes to stopping the spread of misinformation?

Researchers at the Rensselaer Polytechnic Institute (RPI) and the University of Tennessee collaborated to study the role of AI in helping people identify whether the news they’re reading is legitimate or not.

The research paper, “Tailoring Heuristics and Timing AI Interventions for Supporting News Veracity Assessments,” was published in Computers in Human Behavior Reports. It discussed how crowdsourcing marketplace Amazon Mechanical Turk (AMT) can be used to identify misinformation for fresh news and specific heuristics, which are rules of thumb used to process information and consider its veracity. In other words, heuristics are essentially “shortcuts for decisions,” explained Dorit Nevo, an associate professor at RPI’s Lally School of Management and a lead author for the paper.

The study found that AI would be successful in flagging false stories only if the reader did not already have an opinion on the topic, Nevo said. When study subjects were set in their beliefs, confirmation bias kept them from reassessing their views.

Nevo said the first part of the project focused on whether subjects could detect misinformation around climate change and vaccines like the one designed to prevent chicken pox. Then, beginning in April 2020, her team studied how people responded to news related to COVID-19.

“With COVID-19, there was a significant difference,” Nevo said. They found that about 72 percent of respondents could identify misinformation about the coronavirus without heuristic clues, and roughly 93 percent were able to be convinced by the researcher’s heuristics that the content was fake.

Examples of heuristic clues include text with too many capital letters or the use of strong language, Nevo said.

There were two types of heuristics mentioned in the team’s paper: objective heuristics and source heuristics. They put a statement at the top of each article the subjects read; it instructed them to read the article and indicate whether they believed its central thesis.

“We either put a statement that says the AI finds this article reliable and accurate based on the objective heuristics, or we said the AI finds the source reliable,” Nevo said. “So that's the source heuristic.”

In her research on heuristics, Nevo found that people’s thinking takes one of two paths: The first path is to read the article, think about it and decide if they believe it; the second is to consider the source and what others think about the news, and decide whether to believe it before reading it.

Image: Dorit Nevo/RPI/IEEE Spectrum

Researchers at RPI researched the role of heuristics and AI in detecting whether people thought news was credible

Another research paper, “Timing Matters When Correcting Fake News,” published in the Proceedings of the National Academy of Science by researchers at Harvard University, differed from the RPI researchers in its findings. While Nevo and her collaborators found that it’s easier to convince people that a story is fake news before reading it, the Harvard researchers, led by Nadia M. Brashier, a psychologist and neuroscientist, discovered that a fact-check can convince people of misinformation even after reading headlines. When study subjects read true or false labels after reading a headline, that resulted in a 25.3 percent reduction in “subsequent misclassification,” when compared to headlines with no tag, Brashier and her team found.

In the end, fighting misinformation will require both computing and human efforts such as policy changes, says Benjamin D. Horne, an assistant professor of Information Sciences at the University of Tennessee and one of Nevo’s co-authors. He says the RPI-Tennessee work was inspired by AI tools he designed previously. Horne was previously a research assistant at RPI, where he developed machine learning (ML) algorithms that can detect partial truths as well as decontextualized truths and out-of-date information.

“Our algorithms are trained on source-level behavior, both when using the textual content of an article and the network of other news sources that it draws news from,” Horne said. “We have found that these two types of features together are quite good at distinguishing between sources labeled as reliable or unreliable by external news source ratings.”

The machine learning algorithms analyze the writing style and the content-sharing behavior of news outlets, Horne said. Researchers trained a supervised ML algorithm called Random Forest, a classification algorithm that uses decision trees.

AI for Detecting Fake News

So, what’s the potential for AI to be successful in detecting misinformation?

“The tools we have developed, and other tools developed in this area, have fairly high accuracy in lab settings,” says Horne. “For example, our most recent technical work showed around 83% accuracy in predicting when the source of a news article is reliable or unreliable.”

Despite the effectiveness of algorithms, old-fashioned fact-checking by journalists will still be required to combat fake news. AI could filter the information for fact-checkers to verify, according to Horne.

“AI tools are great at dealing with high quantities of information at fast speeds but lack the nuanced analysis that a journalist or fact-checker can provide,” Horne said. “I see a future where the two work together.” Continue reading

Posted in Human Robots

#438731 Video Friday: Perseverance Lands on Mars

Video Friday is your weekly selection of awesome robotics videos, collected by your Automaton bloggers. We’ll also be posting a weekly calendar of upcoming robotics events for the next few months; here's what we have so far (send us your events!):

HRI 2021 – March 8-11, 2021 – [Online Conference]
RoboSoft 2021 – April 12-16, 2021 – [Online Conference]
ICRA 2021 – May 30-5, 2021 – Xi'an, China
Let us know if you have suggestions for next week, and enjoy today's videos.

Hmm, did anything interesting happen in robotics yesterday week?

Obviously, we're going to have tons more on the Mars Rover and Mars Helicopter over the next days, weeks, months, years, and (if JPL's track record has anything to say about it) decades. Meantime, here's what's going to happen over the next day or two:

[ Mars 2020 ]

PLEN hopes you had a happy Valentine's Day!

[ PLEN ]

Unitree dressed up a whole bunch of Laikago quadrupeds to take part in the 2021 Spring Festival Gala in China.

[ Unitree ]

Thanks Xingxing!

Marine iguanas compete for the best nesting sites on the Galapagos Islands. Meanwhile RoboSpy Iguana gets involved in a snot sneezing competition after the marine iguanas return from the sea.

[ Spy in the Wild ]

Tails, it turns out, are useful for almost everything.

[ DART Lab ]

Partnered with MD-TEC, this video demonstrates use of teleoperated robotic arms and virtual reality interface to perform closed suction for self-ventilating tracheostomy patients during COVID -19 outbreak. Use of closed suction is recommended to minimise aerosol generated during this procedure. This robotic method avoids staff exposure to virus to further protect NHS.

[ Extend Robotics ]

Fotokite is a safe, practical way to do local surveillance with a drone.

I just wish they still had a consumer version 🙁

[ Fotokite ]

How to confuse fish.

[ Harvard ]

Army researchers recently expanded their research area for robotics to a site just north of Baltimore. Earlier this year, Army researchers performed the first fully-autonomous tests onsite using an unmanned ground vehicle test bed platform, which serves as the standard baseline configuration for multiple programmatic efforts within the laboratory. As a means to transition from simulation-based testing, the primary purpose of this test event was to capture relevant data in a live, operationally-relevant environment.

[ Army ]

Flexiv's new RIZON 10 robot hopes you had a happy Valentine's Day!

[ Flexiv ]

Thanks Yunfan!

An inchworm-inspired crawling robot (iCrawl) is a 5 DOF robot with two legs; each with an electromagnetic foot to crawl on the metal pipe surfaces. The robot uses a passive foot-cap underneath an electromagnetic foot, enabling it to be a versatile pipe-crawler. The robot has the ability to crawl on the metal pipes of various curvatures in horizontal and vertical directions. The robot can be used as a new robotic solution to assist close inspection outside the pipelines, thus minimizing downtime in the oil and gas industry.

[ Paper ]

Thanks Poramate!

A short film about Robot Wars from Blender Magazine in 1995.

[ YouTube ]

While modern cameras provide machines with a very well-developed sense of vision, robots still lack such a comprehensive solution for their sense of touch. The talk will present examples of why the sense of touch can prove crucial for a wide range of robotic applications, and a tech demo will introduce a novel sensing technology targeting the next generation of soft robotic skins. The prototype of the tactile sensor developed at ETH Zurich exploits the advances in camera technology to reconstruct the forces applied to a soft membrane. This technology has the potential to revolutionize robotic manipulation, human-robot interaction, and prosthetics.

[ ETHZ ]

Thanks Markus!

Quadrupedal robotics has reached a level of performance and maturity that enables some of the most advanced real-world applications with autonomous mobile robots. Driven by excellent research in academia and industry all around the world, a growing number of platforms with different skills target different applications and markets. We have invited a selection of experts with long-standing experience in this vibrant research area

[ IFRR ]

Thanks Fan!

Since January 2020, more than 300 different robots in over 40 countries have been used to cope with some aspect of the impact of the coronavirus pandemic on society. The majority of these robots have been used to support clinical care and public safety, allowing responders to work safely and to handle the surge in infections. This panel will discuss how robots have been successfully used and what is needed, both in terms of fundamental research and policy, for robotics to be prepared for the future emergencies.

[ IFRR ]

At Skydio, we ship autonomous robots that are flown at scale in complex, unknown environments every day. We’ve invested six years of R&D into handling extreme visual scenarios not typically considered by academia nor encountered by cars, ground robots, or AR applications. Drones are commonly in scenes with few or no semantic priors on the environment and must deftly navigate thin objects, extreme lighting, camera artifacts, motion blur, textureless surfaces, vibrations, dirt, smudges, and fog. These challenges are daunting for classical vision, because photometric signals are simply inconsistent. And yet, there is no ground truth for direct supervision of deep networks. We’ll take a detailed look at these issues and how we’ve tackled them to push the state of the art in visual inertial navigation, obstacle avoidance, rapid trajectory planning. We will also cover the new capabilities on top of our core navigation engine to autonomously map complex scenes and capture all surfaces, by performing real-time 3D reconstruction across multiple flights.

[ UPenn ] Continue reading

Posted in Human Robots

#438014 Meet Blueswarm, a Smart School of ...

Anyone who’s seen an undersea nature documentary has marveled at the complex choreography that schooling fish display, a darting, synchronized ballet with a cast of thousands.

Those instinctive movements have inspired researchers at the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), and the Wyss Institute for Biologically Inspired Engineering. The results could improve the performance and dependability of not just underwater robots, but other vehicles that require decentralized locomotion and organization, such as self-driving cars and robotic space exploration.

The fish collective called Blueswarm was created by a team led by Radhika Nagpal, whose lab is a pioneer in self-organizing systems. The oddly adorable robots can sync their movements like biological fish, taking cues from their plastic-bodied neighbors with no external controls required. Nagpal told IEEE Spectrum that this marks a milestone, demonstrating complex 3D behaviors with implicit coordination in underwater robots.

“Insights from this research will help us develop future miniature underwater swarms that can perform environmental monitoring and search in visually-rich but fragile environments like coral reefs,” Nagpal said. “This research also paves a way to better understand fish schools, by synthetically recreating their behavior.”

The research is published in Science Robotics, with Florian Berlinger as first author. Berlinger said the “Bluedot” robots integrate a trio of blue LED lights, a lithium-polymer battery, a pair of cameras, a Raspberry Pi computer and four controllable fins within a 3D-printed hull. The fish-lens cameras detect LED’s of their fellow swimmers, and apply a custom algorithm to calculate distance, direction and heading.

Based on that simple production and detection of LED light, the team proved that Blueswarm could self-organize behaviors, including aggregation, dispersal and circle formation—basically, swimming in a clockwise synchronization. Researchers also simulated a successful search mission, an autonomous Finding Nemo. Using their dispersion algorithm, the robot school spread out until one could detect a red light in the tank. Its blue LEDs then flashed, triggering the aggregation algorithm to gather the school around it. Such a robot swarm might prove valuable in search-and-rescue missions at sea, covering miles of open water and reporting back to its mates.

“Each Bluebot implicitly reacts to its neighbors’ positions,” Berlinger said. The fish—RoboCod, perhaps?—also integrate a Wifi module to allow uploading new behaviors remotely. The lab’s previous efforts include a 1,000-strong army of “Kilobots,” and a robotic construction crew inspired by termites. Both projects operated in two-dimensional space. But a 3D environment like air or water posed a tougher challenge for sensing and movement.

In nature, Berlinger notes, there’s no scaly CEO to direct the school’s movements. Nor do fish communicate their intentions. Instead, so-called “implicit coordination” guides the school’s collective behavior, with individual members executing high-speed moves based on what they see their neighbors doing. That decentralized, autonomous organization has long fascinated scientists, including in robotics.

“In these situations, it really benefits you to have a highly autonomous robot swarm that is self-sufficient. By using implicit rules and 3D visual perception, we were able to create a system with a high degree of autonomy and flexibility underwater where things like GPS and WiFi are not accessible.”

Berlinger adds the research could one day translate to anything that requires decentralized robots, from self-driving cars and Amazon warehouse vehicles to exploration of faraway planets, where poor latency makes it impossible to transmit commands quickly. Today’s semi-autonomous cars face their own technical hurdles in reliably sensing and responding to their complex environments, including when foul weather obscures onboard sensors or road markers, or when they can’t fix position via GPS. An entire subset of autonomous-car research involves vehicle-to-vehicle (V2V) communications that could give cars a hive mind to guide individual or collective decisions— avoiding snarled traffic, driving safely in tight convoys, or taking group evasive action during a crash that’s beyond their sensory range.

“Once we have millions of cars on the road, there can’t be one computer orchestrating all the traffic, making decisions that work for all the cars,” Berlinger said.

The miniature robots could also work long hours in places that are inaccessible to humans and divers, or even large tethered robots. Nagpal said the synthetic swimmers could monitor and collect data on reefs or underwater infrastructure 24/7, and work into tiny places without disturbing fragile equipment or ecosystems.

“If we could be as good as fish in that environment, we could collect information and be non-invasive, in cluttered environments where everything is an obstacle,” Nagpal said. Continue reading

Posted in Human Robots

#437984 WSR: A new Wi-Fi-based system for ...

Researchers at Harvard University have recently devised a system based on Wi-Fi sensing that could enhance the collaboration between robots operating in unmapped environments. This system, presented in a paper pre-published on arXiv, can essentially emulate antenna arrays in the air as a robot moves freely in a 2-D or 3-D environment. Continue reading

Posted in Human Robots