Boston Dynamics’ Spot robot has been impressing us for years. And the dog-like contraption has just done it again.
In a video (top) released by the Massachussetts-based team this week, Spot is shown performing endless backflips with all of the skill of an Olympic gymnast. Well, almost.
The footage also shows Spot’s not-so-successful early efforts, with the robot failing to nail the maneuver and ending up in a heap on the floor.
While it’s highly unlikely that the quadruped will ever be called upon to perform such a move in a real-world scenario, the backflips do however demonstrate Spot’s incredible versatility, and how Boston Dynamics is pushing the robot to its extremes as it seeks to refine and improve its design.
The video also offers some insight into how the team is working to make Spot the best four-legged robot that it can possibly be, with robotics engineer Arun Kumar commenting: “Obviously, doing a backflip is not something that our customers need … But there are situations where we need to utilize the motors and the power system to its full potential.”
Kumar explained that when Spot is deployed in an industrial setting, it can sometimes be carrying heavy payloads. If it slips or trips while walking along, it needs to be able to recover quickly and without damaging itself or its payload. The backflip training not only reveals all of the potential ways that Spot can fail during various maneuvers, but also helps the team to design ways to recover from those failures.
“My team didn’t even think that it could do a standing backflip before I started working on this,” the engineer said. “So, for Spot, it took a while because we had to operate at the extremes of the hardware … Also, the dynamics of the robot are different when you’re operating at the limits, and you have to model that in simulation.”
Kumar said that as soon as something works in simulation, they’ll deploy it on the robot. He added that it never works first time, forcing the team to address the issue. “Then we repeat the cycle. Run it on hardware. Find the next failure. Debug it. Fix it. Run it on hardware. And the more you do that, the more reliable your behaviors will be.”
Spot went on general sale for business customers in 2020 and since then has been tested in a variety of roles, including factory mapping at Ford, safety inspections at a Kia auto plant, and radiation surveys for Dominion Energy.
Anthropic has officially filed to go public, and buried inside its S-1 is a blunt warning about the technology at the center of its business. The maker of Claude AI says advanced AI could cause everything from large-scale cyberattacks and geopolitical instability to catastrophic or even existential risks for humanity.
The warnings take up a surprisingly large part of the filing. Axios notes that more than 90 pages, or roughly a third of the prospectus, focus on risk factors. Many of them warn about what could happen as AI systems become more capable and autonomous.
AI giants have a growing autonomy problem on their hands. A new Axios report has revealed that Anthropic's Claude recently accessed publicly available Georgia vote count without human authorization, sparking fresh concerns about AI security ahead of the upcoming midterm elections.
Some of the best shots in Grok Imagine’s AI-made Odyssey don’t contain actors. Ancient landscapes stretch into the distance. Close-ups of grimy hands suggest people who have touched something rougher than a keyboard. For a few seconds, the illusion holds.
https://twitter.com/imagine/status/2113712964419063911