Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 1,399

Scientists use machine learning to accelerate materials discovery

A new computational approach will improve understanding of different states of carbon and guide the search for materials yet to be discovered.

Materials—we use them, wear them, eat them and create them. Sometimes we invent them by accident, like with Silly Putty. But far more often, making useful materials is a tedious and expensive process of trial and error.

Scientists at the U.S. Department of Energy’s (DOE) Argonne National Laboratory have recently demonstrated an automated process for identifying and exploring promising new materials by combining machine learning (ML)—a type of artificial intelligence—and computing. The new approach could help accelerate the discovery and design of useful materials.

3D Printing With A Drone Swarm?

The goal is to enable the printing of large, complex shaped structures, on any surface, using a swarm of drones, each depositing whatever material is required. It’s a bit like a swarm of wasps building a nest, into whatever little nook they come across, but on the wing.

Even in technical disciplines such as engineering, there is much we can still learn from nature. After all, the endless experimentation and trials of life give rise to some of the most elegant solutions to problems. With that in mind, a large team of researchers took inspiration from the humble (if rather annoying) wasp, specifically its nest-building skills. The idea was to explore 3D printing of structures without the constraints of a framed machine, by mounting an extruder onto a drone.

As you might expect, one of the most obvious issues with this attempt is the tendency of the drone’s to drift around slightly. The solution the team came up with was to mount the effector onto a delta bot carrier hanging from the bottom of the drone, allowing it to compensate for its measured movement and cancel out the majority of the positional error.


The printing method relies upon the use of two kinds of drone. The first done operates as a scanner, measuring the print surface and any printing already completed. The second drone then approaches and lays down a single layer, before they swap places and repeat until the structure is complete.

Watch Google’s Ping-Pong robot pull off a 340-hit rally

As if it weren’t enough to have AI tanning humanity’s hide (figuratively for now) at every board game in existence, Google AI has got one working to destroy us all at Ping-Pong as well. For now they emphasize it is “cooperative,” but at the rate these things improve, it will be taking on pros in no time.

The project, called i-Sim2Real, isn’t just about Ping-Pong but rather about building a robotic system that can work with and around fast-paced and relatively unpredictable human behavior. Ping-Pong, AKA table tennis, has the advantage of being pretty tightly constrained (as opposed to playing basketball or cricket) and a balance of complexity and simplicity.

“Sim2Real” is a way of describing an AI creation process in which a machine learning model is taught what to do in a virtual environment or simulation, then applies that knowledge in the real world. It’s necessary when it could take years of trial and error to arrive at a working model — doing it in a sim allows years of real-time training to happen in a few minutes or hours.

AI material that learns behaviors and adapts to changing conditions

Just like a pianist who learns to play their instrument without looking at the keys or a basketball player who puts in countless hours to throw a seemingly effortless jump shot, UCLA mechanical engineers have designed a new class of material that can learn behaviors over time and develop a “muscle memory” of its own, allowing for real-time adaptation to changing external forces.

The material is composed of a structural system made up of tunable beams that can alter its shape and behaviors in response to dynamic conditions. The research finding, which boasts applications in the construction of buildings, airplanes and imaging technologies among others, was published Wednesday in Science Robotics.

“This research introduces and demonstrates an artificial intelligent material that can learn to exhibit the desired behaviors and properties upon increased exposure to ,” said mechanical and aerospace engineering professor Jonathan Hopkins of the UCLA Samueli School of Engineering who led the research. “The same foundational principles that are used in machine learning are used to give this material its smart and adaptive properties.”

Adobe introduces new collaboration with AI in Photoshop

The new innovative features allow for advanced image editing using artificial intelligence.

Adobe announced new advancements in its Photoshop at its annual Adobe Max conference for technology. These new innovations make the image editing application even smarter in its abilities, and more collaborative. Along with these announcements came a whole new wave of AI advancements and capabilities incorporated into the software.

Adobe.com/Blog.

The flagship desktop app powered by Adobe Sensei AI features numerous improvements, including the one click Delete option and the Fill tool to remove and replace objects with a single click. Along with the AI feature, these improvements were made in time to be introduced at the fall conference. It allows users to remove unwanted elements in their pictures quickly with a shortcut, using Shift + Delete. Another updated feature is the photo restoration neural filter that uses machine learning to detect and get rid of scratches and other small flaws on old photographs.

Google trails ultra-realistic chat tech in the UK that got an engineer fired earlier

‘You’re in a world made of marshmallows!’

A Google app that allows people to communicate with artificial intelligence (AI) systems has been made available in the United Kingdom (U.K.) for a limited trial period.

You’re in a world made of marshmallows! As you take a step, a gentle ‘squish’ comes out under your feet. The marshmallow horizon stretches out in all directions. The sky is a gooey, sticky pink.

Andrey Suslov/iStock.

The app called AI Test Kitchen App made for experimenting with Google’s Language Model for Dialogue Applications (LaMDA), conversational AI, cannot learn new skills from users, although it welcomes comments on how it functions, BBC reported on Wednesday.

Key Immune Cells Classified With New Machine Learning Technique

Researchers from Trinity College Dublin have developed a new, machine learning-based technique to accurately classify the state of macrophages, which are key immune cells. Classifying macrophages is important because they can modify their behaviour and act as pro-or anti-inflammatory agents in the immune response. As a result, the work has a suite of implications for research and has the potential to one day make major societal impact.

For example, this new approach could be of use to drug designers looking to create therapies targeting diseases and auto-immune conditions such as diabetes, cancer and rheumatoid arthritis – all of which are impacted by cellular metabolism and macrophage function.

Because classifying macrophages allows scientists to directly distinguish between macrophage states – based only on their metabolic response under certain conditions – this new information could be used as a diagnosis tool, or to highlight the role of a particular cell type in a disease environment.

A fleet of dog-like robots will soon roam UT Austin’s campus

The project is aided by a $3.6 million grant from the National Science Foundation to the Living and Working With Robots program at UT Austin, under the umbrella of Good Systems, a broad research initiative at the university focused on leveraging the human benefits of AI.

MySA has reached out to UT Austin’s Cockrell School of Engineering for more information on the autonomous robots program.

Keebo optimizes data warehouses with automated ‘learning’ platform, raises $10.5M

Did you miss a session from MetaBeat 2022? Head over to the on-demand library for all of our featured sessions here.

With the rise of Snowflake and other cloud data warehouses, enterprises finally have a simple way to mobilize their data assets at scale. They can easily connect data from different sources and start driving efficiencies while keeping upfront investments (or CapEx) on the lower side.

The benefits of the solutions are unparalleled, but cloud data services also come with the challenge of high operating expenses. Essentially, due to constantly growing datasets, companies have to deal with high compute costs and query performance latencies. Without a solution, their teams have to give about 30–40% of their time to manually develop features that could optimize the warehouse for the required performance and budget constraints.

/* */