Giving robots self-awareness as they move through space - and maybe even providing them with gene-like methods for storing rules of behavior - could be important steps toward creating more intelligent machines.
Now roboticists are going a step further, training their creations to do even better: understand their own image in space and interact with humans like humans do. Today, there are already robot-teachers like KeeKo, robot-pets like Moffin, robot-babysitters like iPal, and robotic companions for the elderly like Pepper.
But even these reasonably intelligent creations still have huge limitations, some scientists think. “There are niche applications for the current generations of robots,” says professor Anthony Zador at Cold Spring Harbor Laboratory—but they are not “generalists” who can do varied tasks all on their own, as they mostly lack the abilities to improvise, make decisions based on a multitude of facts or emotions, and adjust to rapidly changing circumstances. “We don’t have general purpose robots that can interact with the world. We’re ages away from that.”
Robotic spatial self-awareness – the achievement by the team at Columbia – is an important step toward creating more intelligent machines. Hod Lipson, professor of mechanical engineering who runs the Columbia lab, says that future robots will need this ability to assist humans better. Knowing how you look and where in space your parts are, decreases the need for human oversight. It also helps the robot to detect and compensate for damage and keep up with its own wear-and-tear. And it allows robots to realize when something is wrong with them or their parts. “We want our robots to learn and continue to grow their minds and bodies on their own,” Chen says. That’s what Zador wants too—and on a much grander level. “I want a robot who can drive my car, take my dog for a walk and have a conversation with me.”
Columbia scientists have trained a robot to become aware of its own "body," so it can map the right path to touch a ball without running into an obstacle, in this case a square.
Jane Nisselson and Yinuo Qin/ Columbia Engineering
Today’s technological advances are making some of these leaps of progress possible. One of them is the so-called Deep Learning—a method that trains artificial intelligence systems to learn and use information similar to how humans do it. Described as a machine learning method based on neural network architectures with multiple layers of processing units, Deep Learning has been used to successfully teach machines to recognize images, understand speech and even write text.
Trained by Google, one of these language machine learning geniuses, BERT, can finish sentences. Another one called GPT3, designed by San Francisco-based company OpenAI, can write little stories. Yet, both of them still make funny mistakes in their linguistic exercises that even a child wouldn’t. According to a paper published by Stanford’s Center for Research on Foundational Models, BERT seems to not understand the word “not.” When asked to fill in the word after “A robin is a __” it correctly answers “bird.” But try inserting the word “not” into that sentence (“A robin is not a __”) and BERT still completes it the same way. Similarly, in one of its stories, GPT3 wrote that if you mix a spoonful of grape juice into your cranberry juice and drink the concoction, you die. It seems that robots, and artificial intelligence systems in general, are still missing some rudimentary facts of life that humans and animals grasp naturally and effortlessly.
How does one give robots a genome? Zador has an idea. We can’t really equip machines with real biological nucleotide-based genes, but we can mimic the neuronal blueprint those genes create.
It's not exactly the robots’ fault. Compared to humans, and all other organisms that have been around for thousands or millions of years, robots are very new. They are missing out on eons of evolutionary data-building. Animals and humans are born with the ability to do certain things because they are pre-wired in them. Flies know how to fly, fish knows how to swim, cats know how to meow, and babies know how to cry. Yet, flies don’t really learn to fly, fish doesn’t learn to swim, cats don’t learn to meow, and babies don’t learn to cry—they are born able to execute such behaviors because they’re preprogrammed to do so. All that happens thanks to the millions of years of evolutions wired into their respective genomes, which give rise to the brain’s neural networks responsible for these behaviors. Robots are the newbies, missing out on that trove of information, Zador argues.
A neuroscience professor who studies how brain circuitry generates various behaviors, Zador has a different approach to developing the robotic mind. Until their creators figure out a way to imbue the bots with that information, robots will remain quite limited in their abilities. Each model will only be able to do certain things it was programmed to do, but it will never go above and beyond its original code. So Zador argues that we have to start giving robots a genome.
How does one do that? Zador has an idea. We can’t really equip machines with real biological nucleotide-based genes, but we can mimic the neuronal blueprint those genes create. Genomes lay out rules for brain development. Specifically, the genome encodes blueprints for wiring up our nervous system—the details of which neurons are connected, the strength of those connections and other specs that will later hold the information learned throughout life. “Our genomes serve as blueprints for building our nervous system and these blueprints give rise to a human brain, which contains about 100 billion neurons,” Zador says.
If you think what a genome is, he explains, it is essentially a very compact and compressed form of information storage. Conceptually, genomes are similar to CliffsNotes and other study guides. When students read these short summaries, they know about what happened in a book, without actually reading that book. And that’s how we should be designing the next generation of robots if we ever want them to act like humans, Zador says. “We should give them a set of behavioral CliffsNotes, which they can then unwrap into brain-like structures.” Robots that have such brain-like structures will acquire a set of basic rules to generate basic behaviors and use them to learn more complex ones.
Currently Zador is in the process of developing algorithms that function like simple rules that generate such behaviors. “My algorithms would write these CliffsNotes, outlining how to solve a particular problem,” he explains. “And then, the neural networks will use these CliffsNotes to figure out which ones are useful and use them in their behaviors.” That’s how all living beings operate. They use the pre-programmed info from their genetics to adapt to their changing environments and learn what’s necessary to survive and thrive in these settings.
For example, a robot’s neural network could draw from CliffsNotes with “genetic” instructions for how to be aware of its own body or learn to adjust its movements. And other, different sets of CliffsNotes may imbue it with the basics of physical safety or the fundamentals of speech.
At the moment, Zador is working on algorithms that are trying to mimic neuronal blueprints for very simple organisms—such as earthworms, which have only 302 neurons and about 7000 synapses compared to the millions we have. That’s how evolution worked, too—expanding the brains from simple creatures to more complex to the Homo Sapiens. But if it took millions of years to arrive at modern humans, how long would it take scientists to forge a robot with human intelligence? That’s a billion-dollar question. Yet, Zador is optimistic. “My hypotheses is that if you can build simple organisms that can interact with the world, then the higher level functions will not be nearly as challenging as they currently are.”
Recent advancements in engineering mean that the first preclinical trials for an artificial kidney could happen soon.
Still, the devices aren’t ready for testing in humans—yet. But recent advancements in engineering mean that the first preclinical trials for an artificial kidney could happen soon, according to Jonathan Himmelfarb, a nephrologist at the University of Washington.
“It would liberate people with kidney failure,” Himmelfarb says.
An engineering marvel
Compared to the heart or the brain, the kidney doesn’t get as much respect from the medical profession, but its job is far more complex. “It does hundreds of different things,” says UCLA’s Ira Kurtz.
Kurtz would know. He’s worked as a nephrologist for 37 years, devoting his career to helping those with kidney disease. While his colleagues in cardiology and endocrinology have seen major advances in the development of artificial hearts and insulin pumps, little has changed for patients on hemodialysis. The machines remain bulky and require large volumes of a liquid called dialysate to remove toxins from a patient’s blood, along with gallons of purified water. A kidney transplant is the next best thing to someone’s own, functioning organ, but with over 600,000 Americans on dialysis and only about 100,000 kidney transplants each year, most of those in kidney failure are stuck on dialysis.
Part of the lack of progress in artificial kidney design is the sheer complexity of the kidney’s job. Each of the 45 different cell types in the kidney do something different.
Part of the lack of progress in artificial kidney design is the sheer complexity of the kidney’s job. To build an artificial heart, Kurtz says, you basically need to engineer a pump. An artificial pancreas needs to balance blood sugar levels with insulin secretion. While neither of these tasks is simple, they are fairly straightforward. The kidney, on the other hand, does more than get rid of waste products like urea and other toxins. Each of the 45 different cell types in the kidney do something different, helping to regulate electrolytes like sodium, potassium, and phosphorous; maintaining blood pressure and water balance; guiding the body’s hormonal and inflammatory responses; and aiding in the formation of red blood cells.
There's been little progress for patients during Ira Kurtz's 37 years as a nephrologist. Artificial kidneys would change that.
UCLA
Dialysis primarily filters waste, and does so well enough to keep someone alive, but it isn’t a true artificial kidney because it doesn’t perform the kidney’s other jobs, according to Kurtz, such as sensing levels of toxins, wastes, and electrolytes in the blood. Due to the size and water requirements of existing dialysis machines, the equipment isn’t portable. Physicians write a prescription for a certain duration of dialysis and assess how well it’s working with semi-regular blood tests. The process of dialysis itself, however, is conducted blind. Doctors can’t tell how much dialysis a patient needs based on kidney values at the time of treatment, says Meera Harhay, a nephrologist at Drexel University in Philadelphia.
But it’s the impact of dialysis on their day-to-day lives that creates the most problems for patients. Only one-quarter of those on dialysis are able to remain employed (compared to 85% of similar-aged adults), and many report a low quality of life. Having more flexibility in life would make a major different to her patients, Harhay says.
“Almost half their week is taken up by the burden of their treatment. It really eats away at their freedom and their ability to do things that add value to their life,” she says.
Art imitates life
The challenge for artificial kidney designers was how to compress the kidney’s natural functions into a portable, wearable, or implantable device that wouldn’t need constant access to gallons of purified and sterilized water. The other universal challenge they faced was ensuring that any part of the artificial kidney that would come in contact with blood was kept germ-free to prevent infection.
As part of the 2021 KidneyX Prize, a partnership between the U.S. Department of Health and Human Services and the American Society of Nephrology, inventors were challenged to create prototypes for artificial kidneys. Himmelfarb’s team at the University of Washington’s Center for Dialysis Innovation won the prize by focusing on miniaturizing existing technologies to create a portable dialysis machine. The backpack sized AKTIV device (Ambulatory Kidney to Increase Vitality) will recycle dialysate in a closed loop system that removes urea from blood and uses light-based chemical reactions to convert the urea to nitrogen and carbon dioxide, which allows the dialysate to be recirculated.
Himmelfarb says that the AKTIV can be used when at home, work, or traveling, which will give users more flexibility and freedom. “If you had a 30-pound device that you could put in the overhead bins when traveling, you could go visit your grandkids,” he says.
Kurtz’s team at UCLA partnered with the U.S. Kidney Research Corporation and Arkansas University to develop a dialysate-free desktop device (about the size of a small printer) as the first phase of a progression that will he hopes will lead to something small and implantable. Part of the reason for the artificial kidney’s size, Kurtz says, is the number of functions his team are cramming into it. Not only will it filter urea from blood, but it will also use electricity to help regulate electrolyte levels in a process called electrodeionization. Kurtz emphasizes that these additional functions are what makes his design a true artificial kidney instead of just a small dialysis machine.
One version of an artificial kidney.
UCLA
“It doesn't have just a static function. It has a bank of sensors that measure chemicals in the blood and feeds that information back to the device,” Kurtz says.
Other startups are getting in on the game. Nephria Bio, a spinout from the South Korean-based EOFlow, is working to develop a wearable dialysis device, akin to an insulin pump, that uses miniature cartridges with nanomaterial filters to clean blood (Harhay is a scientific advisor to Nephria). Ian Welsford, Nephria’s co-founder and CTO, says that the device’s design means that it can also be used to treat acute kidney injuries in resource-limited settings. These potentials have garnered interest and investment in artificial kidneys from the U.S. Department of Defense.
For his part, Burton is most interested in an implantable device, as that would give him the most freedom. Even having a regular outpatient procedure to change batteries or filters would be a minor inconvenience to him.
“Being plugged into a machine, that’s not mimicking life,” he says.
This article was first published by Leaps.org on May 5, 2022.
New research focuses on methods that could change medicine-making worldwide. The scientists propose bursting cells open, removing their DNA and using the cellular gears inside to make therapies.
Rao and his team of collaborators, which spans multiple research institutions, believe they have a better approach that may change medicine-making worldwide. They suggest forgoing the concept of using living cells as medicine-producers. Instead, they propose breaking the cells and using the remaining cellular gears for assembling the therapeutic compounds. Instead of inserting the DNA into living cells, the team burst them open, and removed their DNA altogether. Yet, the residual molecular machinery of ribosomes, polymerases and other cogwheels still functioned the way it would in a cell. “Now if you drop your DNA drug-making instructions into that soup, this machinery starts making what you need,” Rao explains. “And because you're no longer worrying about living cells, it becomes much simpler and more efficient.” The collaborators detail their cell-free protein synthesis or CFPS method in their recent paper published in preprint BioAxiv.
While CFPS does not use living cells, it still needs the basic building blocks to assemble proteins from—such as amino acids, nucleotides and certain types of enzymes. These are regularly added into this “soup” to keep the molecular factory chugging. “We just mix everything in as a batch and we let it integrate,” says James Robert Swartz, professor of chemical engineering and bioengineering at Stanford University and co-author of the paper. “And we make sure that we provide enough oxygen.” Rao likens the process to making milk from milk powder.
For a variety of reasons—from the field’s general inertia to regulatory approval hurdles—the method hasn’t become mainstream. The pandemic rekindled interest in medicines that can be made quickly and easily, so it drew more attention to the technology.
The idea of a cell-free protein synthesis is older than one might think. Swartz first experimented with it around 1997, when he was a chemical engineer at Genentech. While working on engineering bacteria to make pharmaceuticals, he discovered that there was a limit to what E. coli cells, the workhorse darling of pharma, could do. For example, it couldn’t grow and properly fold some complex proteins. “We tried many genetic engineering approaches, many fermentation, development, and environmental control approaches,” Swartz recalls—to no avail.
“The organism had its own agenda,” he quips. “And because everything was happening within the organism, we just couldn't really change those conditions very easily. Some of them we couldn’t change at all—we didn’t have control.”
It was out of frustration with the defiant bacteria that a new idea took hold. Could the cells be opened instead, so that the protein-forming reactions could be influenced more easily? “Obviously, we’d lose the ability for them to reproduce,” Swartz says. But that also meant that they no longer needed to keep the cells alive and could focus on making the specific reactions happen. “We could take the catalysts, the enzymes, and the more complex catalysts and activate them, make them work together, much as they would in a living cell, but the way we wanted.”
In 1998, Swartz joined Stanford, and began perfecting the biochemistry of the cell-free method, identifying the reactions he wanted to foster and stopping those he didn’t want. He managed to make the idea work, but for a variety of reasons—from the field’s general inertia to regulatory approval hurdles—the method hasn’t become mainstream. The pandemic rekindled interest in medicines that can be made quickly and easily, so it drew more attention to the technology. For their BioArxiv paper, the team tested the method by growing a specific antiviral protein called griffithsin.
First identified by Barry O’Keefe at National Cancer Institute over a decade ago, griffithsin is an antiviral known to interfere with many viruses’ ability to enter cells—including HIV, SARS, SARS-CoV-2, MERS and others. Originally isolated from the red algae Griffithsia, it works differently from antibodies and antibody cocktails.
Most antiviral medicines tend to target the specific receptors that viruses use to gain entry to the cells they infect. For example, SARS-CoV-2 uses the infamous spike protein to latch onto the ACE2 receptor of mammalian cells. The antibodies or other antiviral molecules stick to the spike protein, shutting off its ability to cling onto the ACE2 receptors. Unfortunately, the spike proteins mutate very often, so the medicines lose their potency. On the contrary, griffithsin has the ability to cling to the different parts of viral shells called capsids—namely to the molecules of mannose, a type of sugar. That extra stuff, glued all around the capsid like dead weight, makes it impossible for the virus to squeeze into the cell.
“Every time we have a vaccine or an antibody against a specific SARS-CoV-2 strain, that strain then mutates and so you lose efficacy,” Rao explains. “But griffithsin molecules glom onto the viral capsid, so the capsid essentially becomes a sticky mess and can’t enter the cell.” Mannose molecules also don’t mutate as easily as viruses’ receptors, so griffithsin-based antivirals do not have to be constantly updated. And because mannose molecules are found on many viruses’ capsids, it makes griffithsin “a universal neutralizer,” Rao explains.
“When griffithsin was discovered, we recognized that it held a lot of promise as a potential antiviral agent,” O’Keefe says. In 2010, he published a paper about griffithsin efficacy in neutralizing viruses of the corona family—after the first SARS outbreak in the early 2000s, the scientific community was interested in such antivirals. Yet, griffithsin is still not available as an off-the-shelf product. So during the Covid pandemic, the team experimented with synthesizing griffithsin using the cell-free production method. They were able to generate potent griffithsin in less than 24 hours without having to grow living cells.
The antiviral protein isn't the only type of medicine that can be made cell-free. The proteins needed for vaccine production could also be made the same way. “Such portable, on-demand drug manufacturing platforms can produce antiviral proteins within hours, making them ideal for combating future pandemics,” Rao says. “We would be able to stop the pandemic before it spreads.”
Top: Describes the process used in the study. Bottom: Describes how the new medicines and vaccines could be made at the site of a future viral outbreak.
Image courtesy of Rao and team, sourced from An approach to rapid distributed manufacturing of broad spectrumanti-viral griffithsin using cell-free systems to mitigate pandemics.
Rao’s idea is to perfect the technology to the point that any hospital or pharmacy can load up the media containing molecular factories, mix up the required amino acids, nucleotides and enzymes, and harvest the meds within hours. That will allow making medicines onsite and on demand. “That would be a self-contained production unit, so that you could just ship the production wherever the pandemic is breaking out,” says Swartz.
These units and the meds they produce, will, of course, have to undergo rigorous testing. “The biggest hurdles will be validating these against conventional technology,” Rao says. The biotech industry is risk-averse and prefers the familiar methods. But if this approach works, it may go beyond emergency situations and revolutionize the medicine-making paradigm even outside hospitals and pharmacies. Rao hopes that someday the method might become so mainstream that people may be able to buy and operate such reactors at home. “You can imagine a diabetic patient making insulin that way, or some other drugs,” Rao says. It would work not unlike making baby formula from the mere white powder. Just add water—and some oxygen, too.