Tag Archives: deep

#438801 This AI Thrashes the Hardest Atari Games ...

Posted on March 4, 2021 by Android

Learning from rewards seems like the simplest thing. I make coffee, I sip coffee, I’m happy. My brain registers “brewing coffee” as an action that leads to a reward.

That’s the guiding insight behind deep reinforcement learning, a family of algorithms that famously smashed most of Atari’s gaming catalog and triumphed over humans in strategy games like Go. Here, an AI “agent” explores the game, trying out different actions and registering ones that let it win.

Except it’s not that simple. “Brewing coffee” isn’t one action; it’s a series of actions spanning several minutes, where you’re only rewarded at the very end. By just tasting the final product, how do you learn to fine-tune grind coarseness, water to coffee ratio, brewing temperature, and a gazillion other factors that result in the reward—tasty, perk-me-up coffee?

That’s the problem with “sparse rewards,” which are ironically very abundant in our messy, complex world. We don’t immediately get feedback from our actions—no video-game-style dings or points for just grinding coffee beans—yet somehow we’re able to learn and perform an entire sequence of arm and hand movements while half-asleep.

This week, researchers from UberAI and OpenAI teamed up to bestow this talent on AI.

The trick is to encourage AI agents to “return” to a previous step, one that’s promising for a winning solution. The agent then keeps a record of that state, reloads it, and branches out again to intentionally explore other solutions that may have been left behind on the first go-around. Video gamers are likely familiar with this idea: live, die, reload a saved point, try something else, repeat for a perfect run-through.

The new family of algorithms, appropriately dubbed “Go-Explore,” smashed notoriously difficult Atari games like Montezuma’s Revenge that were previously unsolvable by its AI predecessors, while trouncing human performance along the way.

It’s not just games and digital fun. In a computer simulation of a robotic arm, the team found that installing Go-Explore as its “brain” allowed it to solve a challenging series of actions when given very sparse rewards. Because the overarching idea is so simple, the authors say, it can be adapted and expanded to other real-world problems, such as drug design or language learning.

Growing Pains
How do you reward an algorithm?

Rewards are very hard to craft, the authors say. Take the problem of asking a robot to go to a fridge. A sparse reward will only give the robot “happy points” if it reaches its destination, which is similar to asking a baby, with no concept of space and danger, to crawl through a potential minefield of toys and other obstacles towards a fridge.

“In practice, reinforcement learning works very well, if you have very rich feedback, if you can tell, ‘hey, this move is good, that move is bad, this move is good, that move is bad,’” said study author Joost Huinzinga. However, in situations that offer very little feedback, “rewards can intentionally lead to a dead end. Randomly exploring the space just doesn’t cut it.”

The other extreme is providing denser rewards. In the same robot-to-fridge example, you could frequently reward the bot as it goes along its journey, essentially helping “map out” the exact recipe to success. But that’s troubling as well. Over-holding an AI’s hand could result in an extremely rigid robot that ignores new additions to its path—a pet, for example—leading to dangerous situations. It’s a deceptive AI solution that seems effective in a simple environment, but crashes in the real world.

What we need are AI agents that can tackle both problems, the team said.

Intelligent Exploration
The key is to return to the past.

For AI, motivation usually comes from “exploring new or unusual situations,” said Huizinga. It’s efficient, but comes with significant downsides. For one, the AI agent could prematurely stop going back to promising areas because it thinks it had already found a good solution. For another, it could simply forget a previous decision point because of the mechanics of how it probes the next step in a problem.

For a complex task, the end result is an AI that randomly stumbles around towards a solution while ignoring potentially better ones.

“Detaching from a place that was previously visited after collecting a reward doesn’t work in difficult games, because you might leave out important clues,” Huinzinga explained.

Go-Explore solves these problems with a simple principle: first return, then explore. In essence, the algorithm saves different approaches it previously tried and loads promising save points—once more likely to lead to victory—to explore further.

Digging a bit deeper, the AI stores screen caps from a game. It then analyzes saved points and groups images that look alike as a potential promising “save point” to return to. Rinse and repeat. The AI tries to maximize its final score in the game, and updates its save points when it achieves a new record score. Because Atari doesn’t usually allow people to revisit any random point, the team used an emulator, which is a kind of software that mimics the Atari system but with custom abilities such as saving and reloading at any time.

The trick worked like magic. When pitted against 55 Atari games in the OpenAI gym, now commonly used to benchmark reinforcement learning algorithms, Go-Explore knocked out state-of-the-art AI competitors over 85 percent of the time.

It also crushed games previously unbeatable by AI. Montezuma’s Revenge, for example, requires you to move Pedro, the blocky protagonist, through a labyrinth of underground temples while evading obstacles such as traps and enemies and gathering jewels. One bad jump could derail the path to the next level. It’s a perfect example of sparse rewards: you need a series of good actions to get to the reward—advancing onward.

Go-Explore didn’t just beat all levels of the game, a first for AI. It also scored higher than any previous record for reinforcement learning algorithms at lower levels while toppling the human world record.

Outside a gaming environment, Go-Explore was also able to boost the performance of a simulated robot arm. While it’s easy for humans to follow high-level guidance like “put the cup on this shelf in a cupboard,” robots often need explicit training—from grasping the cup to recognizing a cupboard, moving towards it while avoiding obstacles, and learning motions to not smash the cup when putting it down.

Here, similar to the real world, the digital robot arm was only rewarded when it placed the cup onto the correct shelf, out of four possible shelves. When pitted against another algorithm, Go-Explore quickly figured out the movements needed to place the cup, while its competitor struggled with even reliably picking the cup up.

Combining Forces
By itself, the “first return, then explore” idea behind Go-Explore is already powerful. The team thinks it can do even better.

One idea is to change the mechanics of save points. Rather than reloading saved states through the emulator, it’s possible to train a neural network to do the same, without needing to relaunch a saved state. It’s a potential way to make the AI even smarter, the team said, because it can “learn” to overcome one obstacle once, instead of solving the same problem again and again. The downside? It’s much more computationally intensive.

Another idea is to combine Go-Explore with an alternative form of learning, called “imitation learning.” Here, an AI observes human behavior and mimics it through a series of actions. Combined with Go-Explore, said study author Adrien Ecoffet, this could make more robust robots capable of handling all the complexity and messiness in the real world.

To the team, the implications go far beyond Go-Explore. The concept of “first return, then explore” seems to be especially powerful, suggesting “it may be a fundamental feature of learning in general.” The team said, “Harnessing these insights…may be essential…to create generally intelligent agents.”

Image Credit: Adrien Ecoffet, Joost Huizinga, Joel Lehman, Kenneth O. Stanley, and Jeff Clune Continue reading →

Posted in Human Robots

#438779 Meet Catfish Charlie, the CIA’s ...

Posted on February 27, 2021 by Android

Photo: CIA Museum

CIA roboticists designed Catfish Charlie to take water samples undetected. Why they wanted a spy fish for such a purpose remains classified.

In 1961, Tom Rogers of the Leo Burnett Agency created Charlie the Tuna, a jive-talking cartoon mascot and spokesfish for the StarKist brand. The popular ad campaign ran for several decades, and its catchphrase “Sorry, Charlie” quickly hooked itself in the American lexicon.

When the CIA’s Office of Advanced Technologies and Programs started conducting some fish-focused research in the 1990s, Charlie must have seemed like the perfect code name. Except that the CIA’s Charlie was a catfish. And it was a robot.

More precisely, Charlie was an unmanned underwater vehicle (UUV) designed to surreptitiously collect water samples. Its handler controlled the fish via a line-of-sight radio handset. Not much has been revealed about the fish’s construction except that its body contained a pressure hull, ballast system, and communications system, while its tail housed the propulsion. At 61 centimeters long, Charlie wouldn’t set any biggest-fish records. (Some species of catfish can grow to 2 meters.) Whether Charlie reeled in any useful intel is unknown, as details of its missions are still classified.

For exploring watery environments, nothing beats a robot
The CIA was far from alone in its pursuit of UUVs nor was it the first agency to do so. In the United States, such research began in earnest in the 1950s, with the U.S. Navy’s funding of technology for deep-sea rescue and salvage operations. Other projects looked at sea drones for surveillance and scientific data collection.

Aaron Marburg, a principal electrical and computer engineer who works on UUVs at the University of Washington’s Applied Physics Laboratory, notes that the world’s oceans are largely off-limits to crewed vessels. “The nature of the oceans is that we can only go there with robots,” he told me in a recent Zoom call. To explore those uncharted regions, he said, “we are forced to solve the technical problems and make the robots work.”

Image: Thomas Wells/Applied Physics Laboratory/University of Washington

An oil painting commemorates SPURV, a series of underwater research robots built by the University of Washington’s Applied Physics Lab. In nearly 400 deployments, no SPURVs were lost.

One of the earliest UUVs happens to sit in the hall outside Marburg’s office: the Self-Propelled Underwater Research Vehicle, or SPURV, developed at the applied physics lab beginning in the late ’50s. SPURV’s original purpose was to gather data on the physical properties of the sea, in particular temperature and sound velocity. Unlike Charlie, with its fishy exterior, SPURV had a utilitarian torpedo shape that was more in line with its mission. Just over 3 meters long, it could dive to 3,600 meters, had a top speed of 2.5 m/s, and operated for 5.5 hours on a battery pack. Data was recorded to magnetic tape and later transferred to a photosensitive paper strip recorder or other computer-compatible media and then plotted using an IBM 1130.

Over time, SPURV’s instrumentation grew more capable, and the scope of the project expanded. In one study, for example, SPURV carried a fluorometer to measure the dispersion of dye in the water, to support wake studies. The project was so successful that additional SPURVs were developed, eventually completing nearly 400 missions by the time it ended in 1979.

Working on underwater robots, Marburg says, means balancing technical risks and mission objectives against constraints on funding and other resources. Support for purely speculative research in this area is rare. The goal, then, is to build UUVs that are simple, effective, and reliable. “No one wants to write a report to their funders saying, ‘Sorry, the batteries died, and we lost our million-dollar robot fish in a current,’ ” Marburg says.

A robot fish called SoFi
Since SPURV, there have been many other unmanned underwater vehicles, of various shapes and sizes and for various missions, developed in the United States and elsewhere. UUVs and their autonomous cousins, AUVs, are now routinely used for scientific research, education, and surveillance.

At least a few of these robots have been fish-inspired. In the mid-1990s, for instance, engineers at MIT worked on a RoboTuna, also nicknamed Charlie. Modeled loosely on a blue-fin tuna, it had a propulsion system that mimicked the tail fin of a real fish. This was a big departure from the screws or propellers used on UUVs like SPURV. But this Charlie never swam on its own; it was always tethered to a bank of instruments. The MIT group’s next effort, a RoboPike called Wanda, overcame this limitation and swam freely, but never learned to avoid running into the sides of its tank.

Fast-forward 25 years, and a team from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) unveiled SoFi, a decidedly more fishy robot designed to swim next to real fish without disturbing them. Controlled by a retrofitted Super Nintendo handset, SoFi could dive more than 15 meters, control its own buoyancy, and swim around for up to 40 minutes between battery charges. Noting that SoFi’s creators tested their robot fish in the gorgeous waters off Fiji, IEEE Spectrum’s Evan Ackerman noted, “Part of me is convinced that roboticists take on projects like these…because it’s a great way to justify a trip somewhere exotic.”

SoFi, Wanda, and both Charlies are all examples of biomimetics, a term coined in 1974 to describe the study of biological mechanisms, processes, structures, and substances. Biomimetics looks to nature to inspire design.

Sometimes, the resulting technology proves to be more efficient than its natural counterpart, as Richard James Clapham discovered while researching robotic fish for his Ph.D. at the University of Essex, in England. Under the supervision of robotics expert Huosheng Hu, Clapham studied the swimming motion of Cyprinus carpio, the common carp. He then developed four robots that incorporated carplike swimming, the most capable of which was iSplash-II. When tested under ideal conditions—that is, a tank 5 meters long, 2 meters wide, and 1.5 meters deep—iSpash-II obtained a maximum velocity of 11.6 body lengths per second (or about 3.7 m/s). That’s faster than a real carp, which averages a top velocity of 10 body lengths per second. But iSplash-II fell short of the peak performance of a fish darting quickly to avoid a predator.

Of course, swimming in a test pool or placid lake is one thing; surviving the rough and tumble of a breaking wave is another matter. The latter is something that roboticist Kathryn Daltorio has explored in depth.

Daltorio, an assistant professor at Case Western Reserve University and codirector of the Center for Biologically Inspired Robotics Research there, has studied the movements of cockroaches, earthworms, and crabs for clues on how to build better robots. After watching a crab navigate from the sandy beach to shallow water without being thrown off course by a wave, she was inspired to create an amphibious robot with tapered, curved feet that could dig into the sand. This design allowed her robot to withstand forces up to 138 percent of its body weight.

Photo: Nicole Graf

This robotic crab created by Case Western’s Kathryn Daltorio imitates how real crabs grab the sand to avoid being toppled by waves.

In her designs, Daltorio is following architect Louis Sullivan’s famous maxim: Form follows function. She isn’t trying to imitate the aesthetics of nature—her robot bears only a passing resemblance to a crab—but rather the best functionality. She looks at how animals interact with their environments and steals evolution’s best ideas.

And yet, Daltorio admits, there is also a place for realistic-looking robotic fish, because they can capture the imagination and spark interest in robotics as well as nature. And unlike a hyperrealistic humanoid, a robotic fish is unlikely to fall into the creepiness of the uncanny valley.

In writing this column, I was delighted to come across plenty of recent examples of such robotic fish. Ryomei Engineering, a subsidiary of Mitsubishi Heavy Industries, has developed several: a robo-coelacanth, a robotic gold koi, and a robotic carp. The coelacanth was designed as an educational tool for aquariums, to present a lifelike specimen of a rarely seen fish that is often only known by its fossil record. Meanwhile, engineers at the University of Kitakyushu in Japan created Tai-robot-kun, a credible-looking sea bream. And a team at Evologics, based in Berlin, came up with the BOSS manta ray.

Whatever their official purpose, these nature-inspired robocreatures can inspire us in return. UUVs that open up new and wondrous vistas on the world’s oceans can extend humankind’s ability to explore. We create them, and they enhance us, and that strikes me as a very fair and worthy exchange.

This article appears in the March 2021 print issue as “Catfish, Robot, Swimmer, Spy.”

About the Author
Allison Marsh is an associate professor of history at the University of South Carolina and codirector of the university’s Ann Johnson Institute for Science, Technology & Society. Continue reading →

Posted in Human Robots

#438769 Will Robots Make Good Friends? ...

Posted on February 26, 2021 by Android

In the 2012 film Robot and Frank, the protagonist, a retired cat burglar named Frank, is suffering the early symptoms of dementia. Concerned and guilty, his son buys him a “home robot” that can talk, do household chores like cooking and cleaning, and remind Frank to take his medicine. It’s a robot the likes of which we’re getting closer to building in the real world.

The film follows Frank, who is initially appalled by the idea of living with a robot, as he gradually begins to see the robot as both functionally useful and socially companionable. The film ends with a clear bond between man and machine, such that Frank is protective of the robot when the pair of them run into trouble.

This is, of course, a fictional story, but it challenges us to explore different kinds of human-to-robot bonds. My recent research on human-robot relationships examines this topic in detail, looking beyond sex robots and robot love affairs to examine that most profound and meaningful of relationships: friendship.

My colleague and I identified some potential risks, like the abandonment of human friends for robotic ones, but we also found several scenarios where robotic companionship can constructively augment people’s lives, leading to friendships that are directly comparable to human-to-human relationships.

Philosophy of Friendship
The robotics philosopher John Danaher sets a very high bar for what friendship means. His starting point is the “true” friendship first described by the Greek philosopher Aristotle, which saw an ideal friendship as premised on mutual good will, admiration, and shared values. In these terms, friendship is about a partnership of equals.

Building a robot that can satisfy Aristotle’s criteria is a substantial technical challenge and is some considerable way off, as Danaher himself admits. Robots that may seem to be getting close, such as Hanson Robotics’ Sophia, base their behavior on a library of pre-prepared responses: a humanoid chatbot, rather than a conversational equal. Anyone who’s had a testing back-and-forth with Alexa or Siri will know AI still has some way to go in this regard.

Aristotle also talked about other forms of “imperfect” friendship, such as “utilitarian” and “pleasure” friendships, which are considered inferior to true friendship because they don’t require symmetrical bonding and are often to one party’s unequal benefit. This form of friendship sets a relatively very low bar which some robots, like “sexbots” and robotic pets, clearly already meet.

Artificial Amigos
For some, relating to robots is just a natural extension of relating to other things in our world, like people, pets, and possessions. Psychologists have even observed how people respond naturally and socially towards media artefacts like computers and televisions. Humanoid robots, you’d have thought, are more personable than your home PC.

However, the field of “robot ethics” is far from unanimous on whether we can—or should— develop any form of friendship with robots. For an influential group of UK researchers who charted a set of “ethical principles of robotics,” human-robot “companionship” is an oxymoron, and to market robots as having social capabilities is dishonest and should be treated with caution, if not alarm. For these researchers, wasting emotional energy on entities that can only simulate emotions will always be less rewarding than forming human-to-human bonds.

But people are already developing bonds with basic robots, like vacuum-cleaning and lawn-trimming machines that can be bought for less than the price of a dishwasher. A surprisingly large number of people give these robots pet names—something they don’t do with their dishwashers. Some even take their cleaning robots on holiday.

Other evidence of emotional bonds with robots include the Shinto blessing ceremony for Sony Aibo robot dogs that were dismantled for spare parts, and the squad of US troops who fired a 21-gun salute, and awarded medals, to a bomb-disposal robot named “Boomer” after it was destroyed in action.

These stories, and the psychological evidence we have so far, make clear that we can extend emotional connections to things that are very different to us, even when we know they are manufactured and pre-programmed. But do those connections constitute a friendship comparable to that shared between humans?

True Friendship?
A colleague and I recently reviewed the extensive literature on human-to-human relationships to try to understand how, and if, the concepts we found could apply to bonds we might form with robots. We found evidence that many coveted human-to-human friendships do not in fact live up to Aristotle’s ideal.

We noted a wide range of human-to-human relationships, from relatives and lovers to parents, carers, service providers, and the intense (but unfortunately one-way) relationships we maintain with our celebrity heroes. Few of these relationships could be described as completely equal and, crucially, they are all destined to evolve over time.

All this means that expecting robots to form Aristotelian bonds with us is to set a standard even human relationships fail to live up to. We also observed forms of social connectedness that are rewarding and satisfying and yet are far from the ideal friendship outlined by the Greek philosopher.

We know that social interaction is rewarding in its own right, and something that, as social mammals, humans have a strong need for. It seems probable that relationships with robots could help to address the deep-seated urge we all feel for social connection—like providing physical comfort, emotional support, and enjoyable social exchanges—currently provided by other humans.

Our paper also discussed some potential risks. These arise particularly in settings where interaction with a robot could come to replace interaction with people, or where people are denied a choice as to whether they interact with a person or a robot—in a care setting, for instance.

These are important concerns, but they’re possibilities and not inevitabilities. In the literature we reviewed we actually found evidence of the opposite effect: robots acting to scaffold social interactions with others, acting as ice-breakers in groups, and helping people to improve their social skills or to boost their self-esteem.

It appears likely that, as time progresses, many of us will simply follow Frank’s path towards acceptance: scoffing at first, before settling into the idea that robots can make surprisingly good companions. Our research suggests that’s already happening—though perhaps not in a way of which Aristotle would have approved.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Image Credit: Andy Kelly on Unsplash Continue reading →

Posted in Human Robots

#438731 Video Friday: Perseverance Lands on Mars

Posted on February 20, 2021 by Android

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

#438001 How an Israeli Startup Is Using AI to ...

Posted on January 27, 2021 by Android

The first baby conceived using in-vitro fertilization (IVF) was born in the UK in 1978. Over 40 years later, the technique has become commonplace, but its success rate is still fairly low at around 22 to 30 percent. A female-founded Israeli startup called Embryonics is setting out to change this by using artificial intelligence to screen embryos.

IVF consists of fertilizing a woman’s egg with her partner’s or a donor’s sperm outside of her body, creating an embryo that’s then implanted in the uterus. It’s not an easy process in any sense of the word—physically, emotionally, or financially. Insurance rarely covers IVF, and the costs run anywhere from $12,000 to $25,000 per cycle (a cycle takes about a month and includes stimulating a woman’s ovaries to produce eggs, extracting the eggs, inseminating them outside the body, and implanting an embryo).

Women have to give themselves daily hormone shots to stimulate egg production, and these can cause uncomfortable side effects. After so much stress and expense, it’s disheartening to think that the odds of a successful pregnancy are, at best, one in three.

A crucial factor in whether or not an IVF cycle works—that is, whether the embryo implants in the uterus and begins to develop into a healthy fetus—is the quality of the embryo. Doctors examine embryos through a microscope to determine how many cells they contain and whether they appear healthy, and choose the one that looks most viable.

But the human eye can only see so much, even with the help of a microscope; despite embryologists’ efforts to select the “best” embryo, success rates are still relatively low. “Many decisions are based on gut feeling or personal experience,” said Embryonics founder and CEO Yael Gold-Zamir. “Even if you go to the same IVF center, two experts can give you different opinions on the same embryo.”

This is where Embryonics’ technology comes in. They used 8,789 time-lapse videos of developing embryos to train an algorithm that predicts the likelihood of successful embryo implantation. A little less than half of the embryos from the dataset were graded by embryologists, and implantation data was integrated when it was available (as a binary “successful” or “failed” metric).

The algorithm uses geometric deep learning, a technique that takes a traditional convolutional neural network—which filters input data to create maps of its features, and is most commonly used for image recognition—and applies it to more complex data like 3D objects and graphs. Within days after fertilization, the embryo is still at the blastocyst stage, essentially a microscopic clump of just 200-300 cells; the algorithm uses this deep learning technique to spot and identify patterns in embryo development that human embryologists either wouldn’t see at all, or would require massive collation of data to validate.

On top of the embryo videos, Embryonics’ team incorporated patient data and environmental data from the lab into its algorithm, with encouraging results: the company reports that using its algorithm resulted in a 12 percent increase in positive predictive value (identifying embryos that would lead to implantation and healthy pregnancy) and a 29 percent increase in negative predictive value (identifying embyros that would not result in successful pregnancy) when compared to an external panel of embryologists.

TechCrunch reported last week that in a pilot of 11 women who used Embryonics’ algorithm to select their embryos, 6 are enjoying successful pregnancies, while 5 are still awaiting results.

Embryonics wasn’t the first group to think of using AI to screen embryos; a similar algorithm developed in 2019 by researchers at Weill Cornell Medicine was able to classify the quality of a set of embryo images with 97 percent accuracy. But Embryonics will be one of the first to bring this sort of technology to market. The company is waiting to receive approval from European regulatory bodies to be able to sell the software to fertility clinics in Europe.

Its timing is ripe: as more and more women delay having kids due to lifestyle and career-related factors, demand for IVF is growing, and will likely accelerate in coming years.

The company ultimately hopes to bring its product to the US, as well as to expand its work to include using data to improve hormonal stimulation.

Image Credit: Gerd Altmann from Pixabay Continue reading →

Posted in Human Robots