Startup Shengshu plans to use the money for a “general world model,” paving the way for more practical robot applications.
Category: robotics/AI
New AI method flags fluid flow tipping points before simulations break down
David J. Silvester, a mathematics professor at the University of Manchester, has developed a novel machine-learning method to detect sudden changes in fluid behavior, improving speed and the cost of identifying these instabilities and overcoming one of the major obstacles faced when using machine learning to simulate physical systems. The findings are published in the Journal of Computational Physics.
Computational simulations of mathematical models of fluid flow are essential for everyday applications ranging from predicting the weather to the assessment of nuclear reactor safety. The advent of this simulation capability over the past 50 years has revolutionized the development of fuel-efficient airplanes, and sail configurations on racing yachts can now be optimized in real time, providing the marginal gains needed to win races in the America’s Cup.
Optimized aerodynamics means that modern day cyclists can ride faster, golf balls fly further and Olympic swimmers consistently set world records. Computational fluid dynamics also enables the modeling of the flow of blood in the human heart, making the provision of patient-specific surgery possible.
Compression technique makes AI models leaner and faster while they’re still learning
Training a large artificial intelligence model is expensive, not just in dollars, but in time, energy, and computational resources. Traditionally, obtaining a smaller, faster model either requires training a massive one first and then trimming it down, or training a small one from scratch and accepting weaker performance.
Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), Max Planck Institute for Intelligent Systems, European Laboratory for Learning and Intelligent Systems, ETH, and Liquid AI have now developed a new method that sidesteps this trade-off entirely, compressing models during training, rather than after.
Electrofluidic fiber muscles could enable silent robotic systems
Muscles are remarkably effective systems for generating controlled force, and engineers developing hardware for robots or prosthetics have long struggled to create analogs that can approach their unique combination of strength, rapid response, scalability, and control. But now, researchers at the MIT Media Lab and Politecnico di Bari in Italy have developed artificial muscle fibers that come closer to matching many of these qualities.
Like the fibers that bundle together to form biological muscles, these fibers can be arranged in different configurations to meet the demands of a given task. Unlike conventional robotic actuation systems, they are compliant enough to interface comfortably with the human body and operate silently without motors, external pumps, or other bulky supporting hardware.
The new electrofluidic fiber muscles—electrically driven actuators built in fiber format—are described in a recent paper published in Science Robotics. The work is led by Media Lab Ph.D. candidate Ozgun Kilic Afsar; Vito Cacucciolo, a professor at the Politecnico di Bari; and four co-authors.
AI diffusion models tailor drug molecules to custom-fit protein targets, speeding drug development and evaluation
University of Virginia School of Medicine scientists have developed a bold new approach to drug development and discovery that could dramatically accelerate the creation of new medicines. UVA’s Nikolay V. Dokholyan, Ph.D., and colleagues have developed a suite of artificial intelligence-powered tools, called YuelDesign, YuelPocket and YuelBond, that work together to transform how new drugs are created. The centerpiece, YuelDesign, uses a cutting-edge form of AI called diffusion models to design new drug molecules tailored to fit their protein targets exactly, even accounting for the way proteins flex and shift shape during binding.
A companion tool, YuelPocket, identifies exactly where on a protein a drug can attach, while YuelBond ensures the chemical bonds in designed molecules are accurate. Together, the approach is poised to improve both how new drugs are designed and how quickly and efficiently existing drugs can be evaluated for new purposes.
“Think of it this way: Other methods try to design a key for a lock that’s sitting perfectly still, but in your body, that lock is constantly jiggling and changing shape. Our AI designs the key while the lock is moving, so the fit is much more realistic,” said Dokholyan, of UVA’s Department of Neurology. “This could make a real difference for patients with cancer, neurological disorders and many other conditions where we desperately need better drugs targeting these wiggly proteins but keep hitting dead ends.”
AI-designed proteins built from scratch can recognize specific compounds
Professor Gyu Rie Lee of the Department of Biological Sciences successfully designed artificial proteins that selectively recognize specific compounds using AI through joint research with Professor David Baker. The research, published in the journal Nature Communications, is characterized by using AI to design proteins that recognize specific compounds from scratch (de novo) and implementing them as functional biosensors.
While the conventional approach mainly involved searching for natural proteins or modifying some of their functions, this research is highly significant in that it “custom-built” proteins with desired functions through AI-based design and even completed experimental verification.
In particular, the research team successfully designed a protein that selectively recognizes the stress hormone cortisol and implemented an AI-designed biosensor based on it. This is evaluated as a case that extends beyond protein design to actual measurable sensor technology, solving the long-standing challenge of small-molecule recognition in the field of protein design.
AI can design and run thousands of lab experiments without human hands. Humanity isn’t ready for the new risks this brings to biology
Faster protein engineering could mean faster responses to emerging infections and cheaper drugs.
The dual-use problem
Researchers have raised concerns that these same AI tools could be misused, a challenge known as the dual-use problem: Technologies developed for beneficial purposes can also be repurposed to cause harm.
