Jenga-playing robots, China's new Mach-4 engine, a city-destroying asteroid?

Researchers from the University of California, Berkeley have developed a new robot learning technique that helps them master skills like ‘Jenga whipping’ or building furniture rapidly and accurately.

Combining a mixture of artificial intelligence (AI), reinforcement learning, and human guidance, the new technique allows robots to go from novices to experts in just a few hours. The team at the university’s Sergey Levine’s Robotic AI and Learning Lab developed a new learning technique to help robots learn the basics of a task and then get better at it by practicing in the real world.

As humans learn through trial and error, the reinforcement learning method enables the robot to take feedback from sensors and cameras to discover where it performs well and the points at which it fails. Delve deeper into our Must-Read.

🔥Today’s job of the day as featured on jobs.interestingengineering.com:

> Sr. Azure Middleware Developer - Software Engineer
> Software Engineer

Reading Time: 4 mins

Did a friend forward this e-mail to you? Subscribe here.

MUST-READ

🤖 Robots master Jenga whipping, motherboard assembly with new AI learning method

Postdoctoral fellow Jianlan Luo, who is part of the research team, said the robot required less and less human intervention as it learned from experience. “I needed to babysit the robot for maybe the first 30% or something, and then gradually I could actually pay less attention,” he added in a media statement released by UC Berkeley.

To test their new learning method, the team put its robotic system through various unrelated tasks. The first was ‘Jenga whipping’, which involves using a short cord to “whip” individual Jenga blocks from an assembled tower.

Other tasks included flipping an egg in a frying pan, passing an object from one arm to another, assembling a motherboard from scratch, assembling a car dashboard, and replacing an engine timing belt.

✈️ China tests new Mach 4 turbojet engine, could be used in PLA spy planes

Chinese scientists from the Taihang National Laboratory have completed ground tests on a new superfast turbojet engine. According to reports, the new engine incorporates artificial intelligence and can reach speeds of up to Mach 4.

The engine and research behind it are part of China’s quest to build fast reconnaissance aircraft similar to the US’s legendary Lockheed Martin SR-71 ‘Blackbird.’ Capable of flying at the speed of sound on altitudes of over 85,000 feet, the aircraft made aviation history.

First developed in the 1960s, this iconic ‘Blackbird’ could reach speeds of Mach 3.2 and served with distinction until its final retirement in 1999. Since then, no viable replacement has been developed by any nation, but if reports are accurate, China could be closer than most.

☄️ Football field-sized asteroid with 1.6% chance of hitting Earth could destroy a city

Scientists have revealed that a newly detected football field-sized asteroid has a higher than 1% chance of hitting Earth in the early 2030s. Dubbed a ‘city killer’, an asteroid of this kind could unleash more power than 500 Hiroshima bombs if it did reach our planet.

If such an asteroid were to hit a city, it would be more than big enough to wipe the location off the map effectively. While this sounds apocalyptic, it is essential to note that scientists are not panicking but are watching the asteroid very closely.

“At this point, it’s ‘Let’s pay a lot of attention, let’s get as many assets as we can observing it,'” Bruce Betts, chief scientist of The Planetary Society, told the Associated Free Press (AFP).  

HOT TOPICS OF THE DAY

SCIENCE

> A new study reveals that spinel, a vibrant and rare gem, can also be used as a qubit system for quantum storage. (More)

> Scientists have successfully used 13,000 nuclear spins in a gallium arsenide (GaAs) quantum dot system to create a scalable quantum register. (More)

> Approach based on the quantum anomalous Hall effect does away with the need for applied magnetic fields. (More)

ENERGY

> Researchers at the University of New South Wales (UNSW) have shattered the world record for the efficiency of kesterite solar cells, which is considered a promising alternative to traditional silicon-based solar panels. (More)

> Researchers have concluded that recycling lithium-ion batteries helps recover critical metals. The process is claimed to have lower environmental impacts than mining virgin metals. (More)

> Oshkosh AeroTech group has debuted a series of electric support vehicles for the "Airport of the Future" that includes this driverless EV. (More)

INNOVATION

> Engineers at the University of Hong Kong have developed an advanced aerial robot, akin to a drone, designed for high-speed navigation in unknown environments. (More)

> Headline-hitting DeepSeek R1, a new chatbot by a Chinese startup, has failed abysmally in key safety and security tests conducted by a research team at Cisco in collaboration with researchers from the University of Pennsylvania. (More)

> The Atwater group has made the first experimental measurements of laser-induced motions of miniature lightsails in the lab. (More)

VIDEO

> Even as China proposes the largest dam project in the world, Europe and America are trying to reverse the damage wrought by such structures. But what are the costs of such efforts? And how quickly do we see effective results? (More)

> Bitcoin surged to $90,000 following Trump’s re-election, marking a second cryptocurrency boom during his leadership. (More)

FROM THE WEB

> Does drinking alcohol really take away the blues? It's not what you think

> Are we all aliens? NASA's returned asteroid samples hold the ingredients of life from a watery world

> Aerial robot can navigate unknown environments safely at high speeds

> Medieval toilet reveals lost palace of England’s last Anglo-Saxon king

> Overgrown trees hid 3,800-year-old pyramid-like structure in Peru — until now

> 4,500-year-old warrior cemetery discovered near Förderstedt