D.S.

@chris_j_paxton Found a longer video, they're saying the fridge they mostly shot with was 50 lbs: youtube.com/watch?v=xKK5ze…

@chris_j_paxton It's a prop fridge (light is on even though it's not plugged in). Hard to know how much it weighs. Feels deceptive for BD to call this a feat of strength without revealing those details.

New/mass production atlas carrying an entire mini fridge like it's nothing. Payload capacity for humanoids is often very low, when it really should be one of their defining attributes. As an aside, BD has a long history of these gimmicky one off demos, but they've been churning them out pretty quickly lately. I think that's a good sign that the company is shortening development times and getting closer to something they can ship.

@LeeMcClymont @chrisoffner3d There's a reason it's not handling real packages; it even struggles with the four types of boxes that are available, which is why most of the packages are soft packages. The variety of different packages has been a big issue in automating this in the past, from what I know.

@lswello @chrisoffner3d why? They are not programmed for those packages, they are programmed to identify an object and manipulate it. The variety of packages wont change much

Is anyone checking the false positive rate? I just saw the robot shove a package onto the belt with the barcode facing up. Didn’t see the human make such a mistake.

@mattparlmer Not just factories, either. Lots of effort put into service robotics. For instance, many hotels have extremely useful delivery robots that ride the elevators on their own.

Not only that, but humanoids are a sideshow in Chinese robotics, the main event is the breakneck pace at which they are automating their already exceptionally efficient factories with far more robust robots

@TSoS_ The sequel trilogy crawls are incredibly boring. I remember how curious I was about what happened after Jedi, and then we get TFA's "somehow everything went back to Empire vs. Rebels, I guess." Not to mention none of the sequel crawls actually reflect the movie they're in.

The world-building of this trilogy is so frustrating thanks to J.J. Abrams' inability to evolve the franchise in meaningful ways. Still, TLJ doesn't do much better It is simply not believable for the First Order to take over THE ENTIRE GALAXY in less than a week. Just so boring

@uniquemoviemom Kind of crazy how much this looks like low quality AI slop.

No Jedi with formal training could defeat the strongest Sith in history, so easily.

@TSoS_ And apparently the entirety of the New Republic was a few planets in a single star system.

It's criminal that the only scene the New Republic gets in this ENTIRE TRILOGY is its complete destruction. Laughably bad world-building that seems too afraid to tackle politics because people did not like it in the Prequels. Inherently regressive and boring in a way few things in this franchise were. Convinced this exists only so J.J. Abrams could copy the underdogs vs. an Empire storyline and aesthetic, allowing him to send in 20 X-Wings against a Death Star because he lacks the creativity to make a compelling climax of his own.

@marauder_matt The prequels being poorly thought out doesn't justify the sequels being poorly thought out. It mostly demonstrates the ongoing decline of Star Wars.

Look… the First Order had 30 years to build up arms. The Confederacy if Independent Systems had 10 and was gearing up to build the Death Star… Resource management has never been all that important to Star Wars.

@TSoS_ Maz Kanata felt like Poochie.

I do not like Maz Kanata. I especially do not like Maz Kanata's "a good question for another time" line when she is asked how she got Luke's lightsaber. Did we ever find out how she got it in some comic? I honestly don't know.

@Gunj0n @TSoS_ Finn being in another solar system but seeing the destruction of Hosnian prime in the sky was even dumber. I think that's when I gave up on trying to process the movie, because it was obvious the writers didn't care.

@TSoS_ I still can’t get over the fact that it’s called star killer base and it destroys entire systems but not by killing their star, but by splitting the end of the laser into multiple different beams that strike different planets all at once. The dumbest thing I’ve ever seen

Starkiller base is one of the dumbest things in Star Wars, a 3rd Death Star several times more powerful than the first 2, constructed in secret by an Empire 2.0 that should not have resources of that kind. All so J.J. Abrams could replicate the ANH Trench run (in a worse manner).

@carlosdponx Not being anywhere close to shipping a product is a good indication that they're behind.

One of the major downsides of western humanoid companies not shipping robots yet is they don't get to benefit from people that are good at content just buying their robot and making good content with it. There's a lot of ways they are ahead of chinese companies, but you wouldn't know it if you're not a robotics insider, and media matters in a technology race. There's a lot about the way Brett messages Figure that I don't like, but I think other companies should follow in his example a little more here and just show off the robot. You don't have to do a whole skit or operation, just point a nice camera at the robot doing work with an overlay for some context! You might think it's boring, but the internet is massive, there's an audience for it, and that audience will clip it and share it ad nauseam

@HedgieMarkets This is a staged demo, you'll notice that the same 4 box shapes are used over and over again. The problem with automating this task in the past has been the sheer variety of boxes. This impressive, but still far off from doing this task in a real, not staged, environment.

🦔Figure AI ran a livestream showing its 03 humanoid robot performing package sorting at what the company billed as 30 hours of continuous operation. The actual setup uses three additional robots in the background that swap in roughly every 4 hours, meaning the "30 hours straight" claim is closer to coordinated shift work between four units than continuous runtime by one. The task itself is package induction, flipping shipping labels face-down for downstream scanning at roughly 1,200 packages per hour. Amazon sort centers currently run this work at 3,500 to 5,500 packages per hour using human workers across four induction lanes. Figure's pitch is that humanoid form factor lets the same robot drop into any existing workflow without rebuilding the physical infrastructure around specialized automation. My Take A humanoid robot moving packages a foot and a half on a conveyor belt is genuinely impressive engineering and a terrible business case at the same time. The throughput is roughly a third of what a human worker manages on the same line, the unit cost is six figures before maintenance, and the underlying task could be solved with a $40,000 sensor tunnel that scans labels regardless of orientation. The entire pitch is built on the idea that hyperscale logistics operators will pay a premium to avoid redesigning their workflows, which is a real business but not the revolutionary one Figure is selling. The reason humanoids are getting funded at this valuation is the marketing optic of human form replacing human labor, not the underlying economics. Task-specific automation has been doing this work for two decades and would do it cheaper if the goal were actually productivity. What Figure is offering is a different product, which is the ability to tell investors and operators that you have replaced humans with something that looks like a human, and that visual is doing most of the work in the funding rounds. The technology will eventually justify the cost in narrower domains like elder care or home robotics where the form factor matters, but selling humanoids to Amazon to flip packages is solving a problem that already had a cheaper answer. Hedgie🤗

@chris_j_paxton Is it? Other humanoids are claiming more, but won't be available as products for years. I'm not sure we've even seen prototype humanoids lifting up more than 15kg?

The new H2 looks incredibly cool. 15kg payload is... incredibly low for its size though.

@davidbarbexi @RedBoulderPunch The impression I always got from the movie was that he was well respected, but also a free spirit. He also shares some of the Council's faults. Obi Wan senses the Sith at the beginning of the movie, and Qui Gon tells him to stop getting distracted.

@RedBoulderPunch that's cuz he’s NOT supposed to be the cool renegade Jedi that the Council doesn’t fully trust because he doesn’t fully abide by their dogma. That's fanon. In Lucas' narrative: Qui-Gon is a Jedi who has SOME ideological differences with the Council, who aren't dogmatic at all.

He’s supposed to be the cool renegade Jedi that the Council doesn’t fully trust because he doesn’t fully abide by their dogma Absolutely nothing in the movie itself or Neeson’s performance communicates this whatsoever in a satisfying or interesting manner

@davidpattersonx @GoodmanAric If you watch you'll notice that there are only 4 types of dummy boxes they use over and over. Average person will get a greater variety of boxes delivered in a week. This is a controlled demo using dummy packages.

@GoodmanAric It's the actual task.

Brett Adcock just announced that this 8-hour demo of Figure 03 humanoid robots doing real work will be extended 24/7. It's interesting how much skepticism and disbelief this has generated. People don't believe humanoid robots will replace full human jobs, even when they see it.

@chris_j_paxton @benhylak There are only 4 types of dummy boxes in this Figure demo, from what I see. I'd imagine the variety of boxes and packages in the real world would lead to significantly worse performance for the robot.

@benhylak Its one of these machines: savethepostoffice.com/a-tour-of-the-… pretty common, usually done by humans because of wide range of package shapes and sizes

can someone explain to me what is going on in this video? it's not sorting anything.... just orienting them?

@MarwaEldiwiny Last year it was only sorting the fake packages in a controlled demo environment for an hour. This year it's several hours. In a few more years, we might have a video of robots sorting fake packages in a controlled demo environment for several days!

Didn’t we already see this demo before? I imagined we’d see bots carrying boxes, walking around, charging themselves continuously, maybe even falling and recovering. What’s being shown right now is something almost any robotic arm can do. People are waiting for real demos...

@Dr_Gingerballs And they don't even trust it enough to do it with real packages. You see in the video they keep reusing the same 4 types of dummy boxes, and they have most of the packages being soft bags (you'll notice the robot struggles more with the boxes).

“So what do you do for a living, Fred?” “Well I used to touch packages.” “I don’t understand…” “The packages would come in, and a silo of them would be fed onto a conveyer belt. As they went by, I touched each one.” “I’m not following here.” “Each bag needed to be touched and I touched each one.” “Okay but why?” “Well the labels need to be face down and I touched them and made them all face down. It was a great gig until my job was eliminated.” “Oh no, did they put another camera looking down from above so that no one needed to touch the package?” “No they spent millions of dollars, or 20 years of my salary to train a robot over months to touch the packages as good as me.” “Why not just add another camera?” “Because that’s not enough to trick investors into buying really inconvenient and expensive automation tools.”

@jeffcafe_ @irvinxyz Even with generalization, wheels are going to be preferable than bipedal designs 99.9% of the time. Legs add significant cost, complexity, and danger for very little benefit. These designs are more about making something that looks like a useful robot than making a useful robot.

@irvinxyz True, but the point here is generalization. The existing physical world was designed around the humanoid form factor, when these arrive in force it’ll be an unlock of the “last mile” of physical automation of tasks.

This is cool. But, nothing here requires a humanoid form factor