Data, Info and News of Life and Economy

Daily Archives: June 23, 2022

Chart: U.S. Manufacturing and Services PMI Dropped Below Market Expectations in June 2022

Source : Bloomberg

Charts: Updated US-China Trade War Tariffs

Source : PIIE

Music Video: The Love You Save

China Achieves ‘Brain-scale’ AI with Latest Supercomputer

Anthony Cuthbertson wrote . . . . . . . . .

Computer scientists in China claim to have run an artificial intelligence program using architecture that is as complex as the human brain.

The AI model, named ‘bagualu’ or ‘alchemist’s pot’, was run on the latest generation of the Sunway supercomputer based at the National Supercomputing Center in the eastern province of Jiangsu.

Researchers described it as a “brain-scale” AI model, which could have applications in fields ranging from self-driving vehicles and computer vision, to chemistry and other scientific discoveries.

The Sunway TaihuLight is officially ranked as the fourth most powerful supercomputer in the world, however the researchers claim the latest demonstration puts it on a par with the US Frontier, which currently tops the list.

The Sunway was ranked as the most powerful computer in the world between 2016 and 2018, according to the Top500 list of leading supercomputers, however Chinese institutions no longer submit performance data to the list. China still has 173 supercomputers listed in the Top500 rankings, more than any other country.

The South China Morning Post reported that the bagualu programme ran with 174 trillion parameters, rivalling the number of synapses in the human brain by some estimates.

The publication reported that the Sunway supercomputer has more than 37 million CPU cores, which is four times as many as the Frontier cupercomputer in the US.

It also has nine petabytes of memory, which is the equivalent of more than 2 million HD movies.

One researcher said that its power gave the latest Sunway the ability to perform parallel computing in a way that mimicked human thinking, claiming it was “like eating while watching television”.

The results were presented at the Principles and Practice of Parallel Programming 2022 conference hosted by the US Association for Computing Machinery in April, but were not reported on at the time.

Source : Yahoo!

Ukrainian Combat Robots Join Fight Against Russian Invasion

David Hambling wrote . . . . . . . . .

Ukrainian forces are getting a new helper, a locally made robotic battlefield scout called GNOM (“Gnome”). The small machine will stealthily reconnoiter Russian positions and provide fire support with a machinegun, according to its maker, a company called Temerland that’s based in Zaporizhia. Gnomes versus ‘Orcs’ may sound like fantasy, but the first robots will enter service next week, the company said in a statement.

While drones seem to be ever-present, remotely operated robots, or Uncrewed Ground Vehicles (UGVs), have so far played little part in this conflict. As the battle lines have stabilized, both forces are increasingly using of portable radio-frequency jammers to knock drones out of the sky which may reduce their impact. GNOM offers an alternative, jam-proof way to spy remotely.

Not much larger than a microwave oven and weighing 50 kg (110 pounds), Temerland says GNOM is highly mobile on four large wheels with 4×4 drive and a quiet 5-horsepower electric motor. The current version is armed with a 7.62mm machinegun. U.S. Army research shows that UGVs make stable firing platforms, allowing a remote gunner to hit targets with considerable accuracy.

While most UGVs are radio-controlled, GNOM spools out a reel of fiber-optic cable behind it. Eduard Trotsenko, CEO and owner of Temerland, told me that the tough, wear-resistant cable provides a broadband link which is immune to radio countermeasures.

“Control of GNOM is possible in the most aggressive environment during the operation of the enemy’s electronic warfare equipment,” says Trotsenko.

Also because the operator is not using a radio, they cannot be detected and targeted by artillery, which may happen to drone operators.

“The operator doesn’t deploy a control station with an antenna, and does not unmask his position,” says Trotsenko. “The cable is not visible, and it also does not create thermal radiation that could be seen by a thermal imager.”

Similar arrangements with fiber optics were used for guided missiles in the early 2000s, notably the French Polyphem and U.S. Army EFOG-M, as well as DARPA’s Close Combat Lethal Recon munition, which developed into the Switchblade. They are also used for some tethered drones and also for remotely operated underwater vehicles, but the sort of electronic warfare seen in Ukraine may see a new demand for fiber-optic control for UGVs.

GNOM’s cable gives it a range of 2,000 meters (1.25 miles); if it is broken the vehicle automatically returns to a predetermined location. While it is usually operated by remote control, GNOM clearly has some onboard intelligence and is capable of autonomous navigation. Previous Temerland designs have included advanced neural network and machine learning hardware and software providing a high degree of autonomy, so the company seems to have experience.

Trotsenko says the machinegun allows GNOM to defend itself and also to provide fire support in situations which might be too dangerous for personnel. He notes that other versions of the GNOM can be used for logistics, intelligence gathering, sabotage and engineering. Temerland has previously shown off a cargo carrier GNOM able to bring ammunition or other supplies to the front line, which can also evacuate casualties with the addition of a special trailer.

A more aggressive GNOM delivers TM62 anti-tank mines: Temerland released a YouTube video showing the robot driving underneath an enemy vehicle and detonating. From underneath, the mines’ 7-kg explosive charge will destroy the heaviest tank, but even getting close should be enough to damage a track and immobilize it. (The Australian Army signed a contract for similar kamikaze ground robots last year).

“Work is underway on mobile platforms for transporting mines,” says Trotsenko. “New designs are being tested.”

Previously the company has announced other possible GNOM variants armed with anti-tank missiles or acting as communications relays or drone carriers.

For the meantime, GNOM will be on scouting duty. Temerland developers say that the vehicle is nearly silent and has a low profile. It can be equipped with a 360-degree camera on a telescoping mast to give a detailed view of the surroundings.

Ukraine fields other remote systems, including a sedan armed with a remote-controlled 14.5mm heavy machinegun, but the GNOM will be the first robotic vehicle on the scene. Russia also has military robots, but so far the only units seen in Ukraine are Uran-6 demining robots; the Uran-9 robotic tank, which performed poorly in Syria, has not shown up in this war.

Tactical robots have long been promoted as a way to reduce casualties and keep soldiers out of the line of fire, while maintaining contact with the enemy. GNOM may prove invaluable for getting a close view of Russian forces – and directing artillery fire on to them – without risking Ukrainian lives.

As the war rages in Ukraine, manufacturers are rolling out large numbers UGVs at the Eurosatory 2022 trade show, some larger and seemingly more sophisticated and more expensive than the GNOM. But the success or otherwise of the small Ukrainian robot in action may do more to shape the future of remote warfare than any of them.

Source : Forbes

Watch video at You Tube (0:28 minute) . . . .

Infographic: Migration of the World’s Millionaires

See large image . . . . . .

Source : Visual Capitalist


记者: 伍岳、成欣、魏玉坤 . . . . . . . . .


























Source : 新华网

Read more at 中华人民共和国外交部

全球发展报告 . . . . .