Inside X’s Mission to Make Robots Boring


These creatures target tabletops. One of them would go to a table and meditate for a few seconds to determine if people were seated; if so, it continues until an empty one is found. After a delay of a second — perhaps taking the algorithmic equivalent of a deep breath before the “Let’s do” moment — the robot spun around and spread his arm, stretching the arm over the table to somehow cover the surface with a clear disinfectant. It then withdrew the arm to squeeze the excess fluid into the bucket at its base. The task is done, it goes on, looking for another table to swipe.

People who end their lunch no longer bother to watch. Robots do this for weeks.

Everyday Robots has built more than 100 robots at X’s Mountain View headquarters.

Photo: Michelle Groskopf

No, this is not a desperate attempt to address the labor shortage. This is research Everyday Robots, an X project, Alphabet’s self-styled “moonshot factory”. The café test site is one of dozens on Google’s campus in Mountain View, California, where a small percentage of the company’s many employees have already returned to work. The project hopes to be useful with robots, operating in the wild rather than controlled environments such as factories. After years of development, Everyday Robots is finally sending its robots into the world — or at least outside the X headquarters building — to do the actual work. Enough of a milestone they invited me to observe, two years after Tom Simonite on WIRED final look at the project. At that point, they have robots that sort the garbage into the proper recycling bin. Janitorial services represent the next, if not the last, frontier.

Attack of the squeegee robots: X’s fleet sweeps the tables of a Google cafe.Video: WIRED Staff

Darcy Grinolds leads the day-to-day hardware reliability and design validation team.

Photo: Michelle Groskopf

I’m young, but it’s a serious thing. Everyday Robots try to do two difficult things, a hairy challenge that some doubt if the effort is worth it. The first is to reliably perform the tasks of human assistants. Everyday Robots live on the razor edge Moravec’s paradox, saying it’s easy for computers to do the hard work mentally and devilishly hard to copy the functions of a two-year-old. Everywhere under the umbrella of the Alphabet, robots navigate complex traffic routes, drive cars safer than humans, and become the Go champion. In the Everyday Robots world, overcoming a mundane task, like crossing a chaotic room and opening a hard handle on the door, is like winning the Super Bowl. The activity of wiping the table, for example, is not just swiping — it involves a whole set of actions that lead to it. Get what happens if the path is blocked by someone or something. “The correct answer for the robot is, OK, do I have enough space to move around well around that?” says Darcy Grinolds, who leads the project’s hardware reliability and design validation team. “Or do I have to change myself completely?”

The second difficult thing the project tried to do was to move towards that goal in a more rational way, in terms of economy and efficiency, with a robot in hand than someone who was tired and left ’ y fee.

Open the door to the future robot.

Video: WIRED Staff

Google, and now X, has been heavily pursuing this vision for over a decade. Leading the Everyday Robots team is Norwegian-born engineer Hans Peter Brondmo, an entrepreneur and engineer who joined X in 2015 and must understand a cacophony of robotics acquisitions by former leader Andy Rubin, who left the company. under a cloud of sexual harassment claims. “Hans Peter is not the obvious choice,” said X CEO Astro Teller. “He’s worried about robotics, but he’s the first person to tell you that he’s not a world-class roboticist. I chose him because he’s a world-class entrepreneur who really understands people. And he’s a socialist fur dye — he’s from Norway! ”

In an office he shared a nonfunctional robot arm he built as a teenager, Brondmo explained that creating an effective general-purpose robot has only become possible with recent innovations. progress in machine learning. Engineers use machine learning to train the software to recognize objects and then run millions of simulations to compress weeks of testing into hours. It helped the robots immersed in his lab to really understand their environment, and build that knowledge to put together a toolset that would help solve the inevitable problems of coping in the wild. While Everyday Robots may not be as bright as the dystopian androids in Boston Dynamics videos, they are optimized to get things done. (Alphabet was previously owned by Boston Dynamics, however sold it to 2017.)



Source link

admin

Leave a Reply

Your email address will not be published. Required fields are marked *