The consultants’ verdict on the Tesla Bot: Elon Musk speedruns robotics 101

Elon Musk has lastly unveiled his Optimus robotic prototype – a bipedal machine that the Tesla CEO envisions will at some point be bought as a “common goal” bot that is cheaper than a automotive and equally able to working in factories and operating errands. at residence.

Optimus’ presentation at Tesla’s AI Day was laced with the same old drunken musk of hype and misdirection. But it surely additionally presents roboticists a very admirable have a look at among the machine’s true capabilities. their judgment? That Tesla engineers have completed a tremendous job in such a short while however that Optimus continues to be extra hype than actuality: a analysis robotic that may take years to grow to be something significant.

Earlier than we dive into the consultants’ opinions, although, let’s recap what we noticed on the day.

  • Two robots are proven. The primary, named Bumble C, is a “tough improvement robotic” seen strolling throughout the stage and performing duties in pre-recorded demos akin to fetching watering cans and packing containers. The second robotic is described by Kasturi as “fairly near what’s going to go into manufacturing” and has an exterior cowl. But it surely was solely used as a prop, wheeled round behind the musk and solely proven waving his hand.
  • Musk reiterated his ambition to create a “helpful humanoid robotic as shortly as doable,” which may very well be “made in very excessive quantity, finally tens of millions of items” and which might “value lower than a automotive — lower than $20,000.” He went on to state that the creation of such a machine would enable “a future with out poverty” and “a basic transformation of civilization as we all know it.”

a:hover]:text-black [&>a]:shadow-underline-gray-63 [&>a:hover]:shadow-underline-black text-gray-63″>Picture: Tesla

  • We have got some fundamental specs for Optimus (however, just like the idea bot, it is unclear if that shall be true the subsequent time we get an replace). It weighs 73kg, has 28 levels of freedom in its limbs, and is powered by a 2.3kWh battery that Tesla claims lasts a full day of labor. The mechanical options that stand out are modeled by human arms, and Tesla confirmed how it’s ported to the software program used for self-driving vehicles for navigation and management.
  • Within the Q&A on the finish of the presentation, Musk repeated hypothesis that the bot may very well be “like a pal” at some point and stated prospects would be capable to order one “inside three years, most likely no more than 5 years” (though it isn’t clear if he means companies or people).

So, these are the naked details of what was stated and proven on stage. However what do the consultants say? Listed here are a few of his reactions, from an interview with The Verge and judgments shared on social media.

Christian Hubicki, professor of robotics at Florida State College, offered his thoughts in a Twitter threadnoting, like many others, that the Tesla crew “got here far in a couple of 12 months” however that Optimus’ potential truly proven on stage “appears customary (however not mind-blowing) for humanoids.”

In his thread, Hubicki notes that the Optimus bot appears to make use of a strolling methodology recognized in robotics because the zero second level, or ZMP (though this has not been confirmed by Tesla). This can be a methodology of motion that has been used for many years, put in in well-known bots akin to Honda’s Asimo. (When you watch Optimus, you may undoubtedly see the similarity in how his knees bend in a semi-crouch, and shift his weight cautiously from one leg to the opposite.) Hubicki notes that this methodology of strolling is “fairly secure, however. not mindblowing in 2022 .

He additionally identified that one other large query the presentation did not reply was reliability. “How usually do you fall? You may’t inform from an excellent video – or perhaps a dwell demo,” he stated. Clearly, relating to inserting robots in factories, reliability is an enormous issue, as any downtime within the meeting line can have a big knock-on impact.

Will Jackson, CEO of Engineered Arts, stated The Verge The primary two prototypes of the Optimus bot had been “decidedly missing” with “no novelty” of their design.

“It is extremely related in idea to the Honda Asimo robotic, the event of which is now deserted,” stated Jackson. “The general design is constructed heavy, clumsy and energy inefficient – the hand could be very fundamental, the one redeeming characteristic is the clutch mechanism within the finger actuation. If you wish to know the way far it’s from the motion and capabilities of the human degree, examine final 12 months – a person in a robotic go well with – and precise {hardware} this 12 months.

Perspective view of the robot picking up the watering can.

Jackson praised Tesla’s AI work on the whole and the dedication of the corporate’s engineers, however questioned the thought of ​​constructing humanoid robots for menial labor. “I am shocked that Musk might attraction to an viewers so enamored with the thought of ​​humanoids and fail to acknowledge that his need to work together with robots is a killer app. Does he assume they’re applauding as a result of the world will lastly have humanoid robots that may raise pipes in automotive factories?”

He concluded that the demo confirmed “an unbelievable dwell demonstration of a herculean effort that sadly lacks novelty and creativeness” and that “hopefully we’ll see a course correction in time for subsequent 12 months’s present.”

Jonathan Aitken, a roboticist and instructor on the College of Sheffield in England, advised The Verge that it was an “fascinating demo” and that “the most important factor was the progress they made whereas they had been engaged on issues.” However he additionally famous that there are numerous unanswered questions in regards to the robotic’s capabilities and the precise nature of the demo.

There’s at all times a giant distinction between the recorded demo and the precise work

Aitken identified that the prerecorded video of Optimus confirmed an “umbilical wire” connected to the robotic, for instance. “That to me asks two large questions, energy and communication/management.” Is it there for safety? Or for energy and command? (It was famous on stage that it was the primary time the robotic had walked and not using a tether.)

“From the presentation, it was good to see the vary of actions that these robots are utilized in,” says Aitken. “Though they’re comparatively restricted.” He famous that the footage of the robotic transferring parts within the Tesla manufacturing unit workstation was spectacular however lacked the context that may clarify how significant the work was. “I want to see extra details about how large the parts are and the way frequent they’re within the manufacturing course of.”

The image shows a humanoid robot walking and holding a box marked with the Tesla logo.

a:hover]:text-black [&>a]:shadow-underline-gray-63 [&>a:hover]:shadow-underline-black text-gray-63″>Picture: Tesla

Cynthia Yeung, head of product at robotic logistics agency Plus One Robotics, live-tweeted his thoughts about the presentationquestioning the corporate’s deal with copying the human kind so intently with it The future business model is planned (which Musk says will contain promoting robots outright as a substitute of promoting their companies like many robotic startups do).

“I believe @elonmusk appears to love the @BostonDynamics strategy to robots (kind over perform) versus what loads of different folks use (perform informs kind),” Yeung’s tweet.

Yeung stated the presentation targeted so much on how the crew had made Optimus stroll however that was it solving these challenges will not necessarily solve real world problems. She additionally identified that the robotic’s five-fingered hand isn’t essentially nearly as good as easy two or three-finger pinchers or vacuum programs. “There is a purpose why all warehouse startups do not use hand-like manipulation mechanisms,” tweeted Yeung.

Elsewhere, Yeung praise the musk to share the highlight along with his crew of engineers saying that the corporate makes some “cool” actuators and simulators. However his general judgment in technical efficiency was damning. “None of that is subtle. Rent some PhDs and go to some robotics conferences,” Yeung’s tweet.

Speak to you The Verge through electronic mail, robotics professor Henrik Christensen from the College of California San Diego stated the mechanical design and strolling expertise of Optimus had been “strong” however that “there’s little innovation” in Tesla’s strategy.

“There is no such thing as a actual proof that it could do fundamental navigation, seize, manipulation.”

Like Yeung, Christensen means that Optimus’ five-fingered hand design appears pointless (“unsure you want 5 fingers … I do not see a robust argument right here past hand waving”). And he famous that a lot was lacking from Tesla’s presentation: “There is no such thing as a actual proof that it could do fundamental navigation, seize, manipulation.”

“It is an incredible early design and it is wonderful how far they’ve are available in 9 months, however innovation past Boston Dynamics and Agility could be very restricted,” Christensen stated. “To this point they’ve constructed an incredible platform to do analysis […] The excellent news is that Musk is not afraid to assume large and make investments. As such I hope to see actual innovation right here sooner or later.

A screenshot from Tesla's presentation shows the robotic arm of the Optimus robot.

a:hover]:text-black [&>a]:shadow-underline-gray-63 [&>a:hover]:shadow-underline-black text-gray-63″>Picture: Tesla

Animesh GargAssistant professor in AI on the College of Toronto, provided a response that focuses very a lot on the broader challenges going through the sector – that’s, why develop the overall goal of humanoid robots within the first place when easy programs are extra achievable?

Writing in a publish shared on Google Docs, Garg stated that creating humanoid bots is a “zero-to-one drawback with no income till the system can ship worth.” That makes investing difficult. Nevertheless, he compares the strategy to designing a completely autonomous automotive and notes that if the ultimate objective could be achieved (clean integration with the human setting), then we save so much by not having to retool the infrastructure to accommodate machines (whether or not they’re robots or vehicles). ).

Garg has some gentle criticisms of the Optimus’ design, noting, for instance, that the cable-driven arm appears to have respectable load capability however gradual response instances. However he’s extra constructive that Musk and Tesla are concerned on this area:

Total, the present design is an efficient first step. Curiosity in constructing such a system is welcome as a result of Tesla and Elon Musk’s involvement within the matter has introduced consideration, expertise, and assets to the matter, setting in movement a flywheel of progress. This effort ought to be praised with cautious optimism by the general public, for the compass to level in the correct path, and Elon brings with him the load of Tesla engineers as we undergo the AI ​​/ Robotics forest.

The image of the Tesla robot is supported on stage by three engineers.

a:hover]:text-black [&>a]:shadow-underline-gray-63 [&>a:hover]:shadow-underline-black text-gray-63″>Picture: Tesla

About the author


Leave a Comment