March 19, 2019

In the middle of the U.S., winds sweeping across the Great Plains generate electricity with such ease that the whole process looks simple. But efficiently shipping that electricity over long distances to cities is a different story. To accommodate the rise of renewables, planners are thinking about switching parts of the grid from alternating current to a more efficient system that uses direct current. That conversion would require rethinking many crucial grid components, including heavy-duty circuit breakers that can handle 100,000 volts. Scientists at GE Research find the challenge positively electrifying.

Breaker, breaker: The Energy Department's R&D arm, ARPA-E, recently awarded $5.8 million to GE Research to develop a superfast medium-voltage direct current circuit breaker, a key component of any DC grid that prevents disruptions from rippling through the system. GE engineers are using charged gas, or plasma, to build a compact breaker that can be adapted to the current infrastructure with as little disruption as possible. Says GE's Timothy Sommerer: 'Nobody wants to run bulldozers through cities to put in new cables and towers and other infrastructure. With a DC grid, you could reuse existing space for AC lines and just upgrade them.'

Learn more here about how GE is wiring the electrical grid of the future.

'Legacy. What is a legacy?' asks Alexander Hamilton in the now-legendary musical, 'Hamilton.' 'It's planting seeds in a garden you never get to see.' What if the founding father was wrong, and the fruits of your 'garden' someday save your life? That's precisely what happened to Dr. Ernie Garcia, a professor of radiology and the director of the Nuclear Cardiology R&D Laboratory at Emory University in Atlanta. When the then-60-year-old researcher experienced chest pains, his own invention - the Emory Cardiac Toolbox (ECTb) - discovered that Garcia had a large coronary artery perfusion defect associated with ischemia. He underwent quadruple bypass surgery and - 11 years later - can regale others with how he became one of the 4 million patients to benefit from ECTb.

Taking the legacy live: Though doctors have been able to view images of a beating heart via a noninvasive test on a PET scanner since the 1970s, they lacked instruments to explain what those images meant, leaving diagnosis open to the art of interpretation. In 1998, Garcia turned that art into science by creating ECTb software - an algorithm that can calculate data from PET scans to determine the patient's likelihood of suffering a heart attack. ECTb is now used in nearly half of all nuclear medicine labs nationwide. In June, GE Healthcare launched a new version of ECTb with Garcia's company Syntermed Live, which allows clinicians to remotely access and review clinical images from virtually any computer system. That could mean saving the lives of patients in small to medium clinics with limited resources. Apparently, Garcia's garden is still growing.

Find more here about how ECTb works to diagnose heart health all over the world.

Love it or hate it, artificial intelligence is increasingly coming to resemble any other co-worker you might interact with on the factory floor or in the office. And so - like with the colleague who rewarms fish leftovers in the office microwave - we'd better figure out how best to work together, writes Davide Valeriani in a new article at The Conversation. Valeriani, a post-doc fellow at Harvard, has been researching how human and AI labor can be complementary. He and his colleague Riccardo Poli looked specifically at security work, investigating whether automation might be able to assist a weary guard tasked with monitoring surveillance footage, as at an airport.

Keeping an A-Eye out: 'It seems like a straightforward request, but it is actually really hard to do,' Valeriani writes - the guard must watch several cameras at a time for hours on end. Facial recognition algorithms can help, but they make mistakes too. Can the human and AI systems work in tandem here? The researchers showed crowd photos to human subjects and algorithms, asking them afterward if they'd seen a particular face and measuring the confidence with which they answered. It turned out humans alone didn't do best, nor did the algorithms on their own - it was a combination of the two. Successful collaborative tasks such as these might address some outstanding ethical concerns over automation, Valeriani writes, as well as 'legal questions about accountability.'

Find out morehere. And clickhere to learn how GE is developing 'humble AI,' a way to make algorithms more people-friendly by imbuing them with a sense of their limits.

1. Supersonic Travel, Hold The Sonic
In its quest to develop a fast-flying jet that doesn't create an earth-shattering boom whenever it breaks the speed of sound, NASA has been able to take some pretty amazing pictures of supersonic shock waves.

2. Black Hole Fun
Theoretically, highly advanced aliens might be able to create tiny, artificial black holes to power their spaceships - and if that's true, humans could use high-powered telescopes to detect their activity, argues a new paper from a Kansas State University mathematician.

3. Read The Lightning
Despite claims to the contrary by rockers, you can't ride a storm - but you could harness a lightning's electromagnetic signals to independently authenticate the security system at your electrical substation, according to researchers at Georgia Tech.

Read more about this week's Coolest Things on Earth here.

- QUOTE OF THE DAY -

'When I got off the machine, I just looked at the computer, and from about 20 feet away I knew exactly what was going on with me.'

- Dr. Ernie Garcia, professor of radiology and the director of the Nuclear Cardiology R&D Laboratory at Emory University

Quote: GE Reports. Image: Getty Images.

ENJOY THIS NEWSLETTER?
Please send it to your friends and let them know they can subscribe here.

Attachments

  • Original document
  • Permalink

Disclaimer

GE - General Electric Company published this content on 19 March 2019 and is solely responsible for the information contained herein. Distributed by Public, unedited and unaltered, on 19 March 2019 09:19:02 UTC