Do Robots Want Legs? What If You Gave ChatGPT a Physique?

Video Friday is your weekly number of superior robotics movies, collected by your folks at IEEE Spectrum robotics. We additionally publish a weekly calendar of upcoming robotics occasions for the following few months. Please ship us your occasions for inclusion.

RSS 2026: 13–17 July 2026, SYDNEY

Summer season College on Multi-Robotic Techniques: 29 July–4 August 2026, PRAGUE

Actuate 2026: 18–19 August 2026, SAN FRANCISCO

IROS 2026: 27 September–1 October 2026, PITTSBURGH

Take pleasure in as we speak’s movies!

Eno is our first agentic robotic: an AI agent and a general-purpose robotic working as one system. It causes, plans, and acts in the true world. Human in functionality, not in type. Each element with a objective, decreased to what issues. Designed to not resemble us, however to increase us. Eno is constructed finish to finish at Genesis.

[ Genesis ]

Engineers from NASA’s Jet Propulsion Laboratory are field-testing superior capabilities for potential future Moon and Mars rovers. Within the Colorado Desert close to Plaster Metropolis, California, groups used a prototype rover known as ERNEST (Exploration Rover for Navigating Excessive Sloped Terrain) to check software program for a possible future long-range lunar mission. The software program allows the rover, developed at JPL, to function autonomously and journey excessive distances with minimal intervention from human operators.

ERNEST is much more succesful than it could look; right here’s some latest analysis exhibiting the sorts of terrain it will possibly deal with:

[ NASA’s Jet Propulsion Lab ]

Desk tennis can produce moments which are tough even for knowledgeable gamers to anticipate…like when the ball clips the web and all of a sudden adjustments path. For the Ace analysis challenge at Sony AI, these occasions had been a key take a look at of the system’s potential to function reliably in unpredictable real-world circumstances. Ace addresses this uncertainty by simulating counterfactual ball trajectories in actual time. Within the video, the inexperienced overlays present these various paths the system considers whereas planning its response.

And take a look at a few of these rallies that the robotic has with Miyuu Khiara.

[ Sony AI ]

This video of an ANYmal deployment in a concrete plant is price watching as a result of it makes express how quadrupeds earn cash in inspection contexts: Amongst different issues, “a cracked crusher basis [was] caught earlier than a week-long shutdown, avoiding roughly $630,000 in misplaced manufacturing.” That pays for lots of robots.

[ ANYbotics ]

A number of fascinating footage right here from GITAI’s prep for a robotic satellite tv for pc servicing demo mission. The thruster test-firing isn’t a robotic, precisely, however it could be the best half.

[ GITAI ]

Anybody who’s tried to take a half first rate picture underwater is aware of that it’s principally not possible, so let’s try to educate robots to manage.

[ Bi-AQUA ]

Thanks, Masato!

Dealing with delicate, irregular or unpredictable objects is among the hardest issues left in automation, and one of the crucial essential. It’s what’s holding again the following wave of robots from doing extra in the true world. That’s why we’re working with PSYONIC on a brand new method. Their Capability Hand, worn by a whole bunch of individuals day by day, captures real-world information on contact, strain and grip. Our GoFa cobot brings the industrial-grade accuracy and repeatability to show that human information into dependable robotic efficiency.

[ ABB Robotics ]

Sanctuary AI has achieved world-class efficiency on a fancy wire-plugging manufacturing activity with a worldwide Tier 1 automotive provider. On this demonstration, Sanctuary AI’s Bodily AI efficiently performs a high-speed wire-plug insertion activity, attaining a validated activity success charge of over 99.5% with a cycle time of simply 2.54 seconds, assembly reside manufacturing benchmarks established by the shopper.

WHY IS THIS STRESSING ME OUT SO MUCH?

[ Sanctuary ]

This video is sort of clearly faux, however I suppose possibly there’s a marketplace for further beefy quadrupeds? Possibly?

[ Kepler ]

I can’t overstate how a lot I don’t want any robotic to take a look at what I’m sporting after which try and promote me issues based mostly on what it thinks it will possibly guess about my character or pursuits.

[ MagicLab ]

I’m right here for fed-up robots studying how one can transfer packing containers by simply kicking them.

[ ATARI Lab ]

Ah, sure, very helpful and crucial robots that make me very uncomfortable.

[ Paper ]

The thought right here is to construct an open course round this challenge, Brit says, “so everybody can expertise bodily AI proper now in a low-risk means.”

[ GrowBot ]

Thanks, Brit!

From Your Website Articles

Associated Articles Across the Net