Connect with us

General

This Robot Just Wants a Hug

I love hugs. Like, really love them. I’d choose a warm embrace over a cream cheese-smeared bagel, an Unbreakable Kimmy Schmidt binge, or a sale on scented candles.

So you can bet I’m excited about the prospect of a humanoid robot programmed to hug people.

HuggieBot—a modified Willow Garage PR2 robot—is the first step toward an IRL Baymax: As tall as an average human, it is made with layers of foam, polyester, and other materials for extra-soft comfort.

(And, as an added bonus, it won’t deflate when the battery runs low.)

“We’re interested in enabling robots to hug because of how common hugs are in daily life and because of their numerous health benefits,” according to lead researcher Alexis Block, a Ph.D. student in the Haptic Intelligence Department at the Max Planck Institute for Intelligent Systems.

Studies suggest that hugs—whether between family, friends, or significant others—can ease stress, lower blood pressure, and make us feel supported, all of which helps stave off infection.

[embedded content]

Earlier this year, Block presented her own findings, based on a study in which HuggieBot gave 30 participants 12 different mechanical hugs.

The experiment, conducted at the University of Pennsylvania, featured the cyborg raising its arms expectantly and asking, “Can I have a hug, please?”

Some folks were wary of embracing the 450-pound computer, while others seemed excited for the novel opportunity.

No one was crushed to death or ran away screaming, NBC News MACH reported; some participants told researchers that HuggieBot was “nicer to hug” than they anticipated.

Researchers are working on a second-generation machine—one that measures how much emotional support its hugs provide. Their ultimate goal, according to MACH, is to “demonstrate that robots are able to comfort people” young and old.

Block has no intention, though, of replacing human embraces (because who would want to?). She simply wants to supplement them.

“We’re advocating for this technology to be used as a complement to other people,” she said, “in situations where it is difficult or uncomfortable to get the support a person needs or wants from another human”—like a college campus or senior living facility.

Let us know what you like about Geek by taking our survey.

SOURCE

General

AI KOs Pro DotA 2 Players in Live Tournament

In another win for artificial intelligence, AI bots successfully defeated a group of professional Defense of the Ancients (DotA) 2 players.

The multiplayer online battle arena mod pits two teams against each other in an attempt to destroy their opponent’s home base, known as an Ancient.

And that’s exactly what OpenAI Five—a set of five cooperative machine learning systems—did during a recent tournament.

The day began with a warm-up: audience volunteers playing the first public match against Five, which won in 14 minutes (an evenly matched game generally takes 45 minutes).

Once limbered up, the AI unit took on—and wiped the floor with—five North American pros: William “Blitz” Lee, Austin “Capitalist” Walsh, Ioannis “Fogged” Loucas, Ben “Merlini” Wu, and David “MoonMeander” Tan.

[embedded content]

Boasting a new ability to draft heroes, OpenAI Five won the first game in only 21 minutes and 37 seconds, and the second in fewer than 25 minutes.

This victory “is a step towards advanced AI systems which can handle the complexity and uncertainty of the real world,” OpenAI wrote in a blog announcement.

For the third match, the non-profit relinquished its greatest skill, instead of allowing audience members to select Five’s characters, putting the machine at a severe disadvantage. It ultimately lost to the humans after 35 minutes and 47 seconds.

“These results,” according to the blog, “give us confidence in moving to the next phase of this project: playing a team of professionals at The International,” held in Vancouver from Aug. 20-25.

Keep an eye on social media for additional game details

But OpenAI Five has dreams greater than DotA 2.

“Ultimately, we will measure the success of our DotA system in its application to real-world tasks,” the firm said.

Founded in 2015 by Elon Musk and Sam Altman, OpenAI aims to promote and develop friendly AI; it also collaborates with other institutions and researchers by making patents and research open to the public.

For all the latest in eSports news head over to our section here.

Let us know what you like about Geek by taking our survey.

SOURCE

Continue Reading

General

Students Build Shoelace-Tying Robot for $600

Over, under, around, and through: Students at the University of California, Davis, engineered a robot capable of tying a shoe.

And it cost far less than any state-of-the-art android.

“This machine,” according to a video published by team member Andrew Choi, “was designed and manufactured with the limitations of only being able to use two motors and a $600 budget.”

Neither compact nor speedy, the device, which uses the Ian Knot (“world’s fastest,” according to its creator), is certainly not going to be part of IKEA’s winter catalogue.

And while DARPA could probably build something faster and sleeker—that also climbs stairs and pulls people from burning wreckage—this contraption is clever, innovative, and, perhaps most importantly: cheap.

[embedded content]

(It remains unclear, however, whether the machine actually works with a foot placed inside the shoe.)

Initially reported on Reddit—posted by UC Davis lecturer Jason Moore (moorepants)—the shoe-tying robot has received a lot of attention since its construction for an annual design competition with Meijo University in Japan.

“We, the professors, come up with machine design challenges,” Moore explained in a statement to Geek. “Student groups in Meijo and UCD work on the machines for about five months independently, and then they come together in Davis to compete.

“The challenges are designed to test the students’ ingenuity, let them make use of their new engineering skills, and to help them learn some about how culture affects machine design,” he continued. “This group did excellent work. It is the only fully functioning shoe typing machine we’ve been able to find on the Internet.”

If this is what a group of five novice engineers can do with $600 and two motors, imagine the possibilities given more money and equipment.

The team includes Choi, Gabriela Gomes, Jacklyn Tran, Stephanie Thai, and Joel Humes.

UC Davis has a history of interesting robots: Biologist Gail Patricelli, of the Department of Evolution and Ecology, recently developed robotic fowl, capable of flirting its way into the hearts of male sage-grouses. The study aimed to learn courtship tactics and analyze coupling decisions.

Editor’s note: This article was updated at 11:50 a.m. ET with comment from Jason Moore.

[embedded content]

Let us know what you like about Geek by taking our survey.

SOURCE

Continue Reading

General

AI Tool Quicksilver Fills Gaps in Wikipedia Knowledge

It’s not exactly news that women are overlooked and underappreciated.

But thanks to a new machine learning system, female scientists, technologists, engineers, and mathematicians (among other professionals) will finally get their digital due.

San Francisco-based Primer is using artificial intelligence to fill in existing gaps in human-generated knowledge—starting with notable ladies like Canadian roboticist Joelle Pineau, addiction treatment researcher Miriam Adelson, and Evelyn Wang, head of MIT’s MechE department.

None of whom, until this month, had articles on English Wikipedia. Yet, someone took the time to curate a “list of people who have lived at airports” for the online encyclopedia.

“I didn’t discover those people on my own. I used a machine learning system we’re building,” John Bohannon, director of science at Primer, wrote in a recent blog post. “It discovered and described them for me.”

(And the Internet created entries for him.)

“It does this much as a human would, if a human could read 500 million news articles, 39 million scientific papers, all of Wikipedia, and then write 70,000 biographical summaries of scientists,” Bohannon added.

It’s called Quicksilver. And it’s an homage to Neal Stephenson’s Baroque Cycle novel series, which imagines technology that captures all human information “in a vast encyclopedia that will be a sort of machine, not only for finding old knowledge but for making new.”

Trained on 30,000 English Wikipedia articles about scientists, their Wikidata entries, and more than 3 million sentences from news documents, as well as the names and affiliations of 200,000 authors of scientific papers, Quicksilver produced shocking results.

“In the morning we found 40,000 people missing from Wikipedia who have a similar distribution of news coverage as those who do have articles,” Bohannon said. “Quicksilver doubled the number of scientists potentially eligible for a Wikipedia article overnight.”

Launched more than 17 years ago, Wikipedia remains one of the largest and most popular general reference works on the Internet. But it takes a lot of upkeep; every article must be meticulously cited and regularly updated—forever.

Despite a decade of attempts, no one has managed to crack the code for computer-generated passages. And Primer doesn’t plan to, either.

Rather than use the World Wide Web “as an academic testbed for summarization algorithms,” Bohannon said, the firm is working on a system “that can be used for building and maintaining knowledge bases” like Wikipedia.

As an experiment, Primer is publishing a sample of 100 short Quicksilver-generated summaries of scientists missing from Wikipedia.

“We’re curious how long it will take before someone creates their articles,” Bohannon said.

In the span of about a week, all three women mentioned in his blog have since been covered on the site. Three down, tens of thousands to go.

To learn more about AI click here.

[embedded content]

Let us know what you like about Geek by taking our survey.

SOURCE

Continue Reading

Deals

Trending