[Tiereven]

As we enter an era of wide scale robotic deployment, we need to think long and hard about what the maintenance bottleneck will look like. We need to advocate now for reliable and open upgrades, replacement parts, service documentation, and diagnostics.

Right to repair will be even more important for this technology than autos or general computing.

[LeifCarrotson]

Literally standing in front of a proprietary Fanuc industrial 6-axis arm waiting for Roboguide at the moment... this is already a wide scale industry and shows low probability to trend towards open and repairable technology.

There have been some efforts for vendor-agnostic robot software like RoboDK and other warehouse execution systems, but the default is proprietary vendor software.

It would be nice for society if this were true, but we'd need someone to exist whose complementary technology was robotics who found it worth commoditizing the entire ecosystem against their will. Or regulators who weren't entirely beholden to industry lobbyists.

[delfinom]

Personal experience.

Fanuc robots are straight forward to service, they make the parts very available to do it yourself if you want. We order them here and there no problem.

But they are beasts and it can take an entire day just to replace a part. Then you have to reassemble it in the right order. None of it is made difficult on purpose. It has tight tolerances, and fancy shit like harmonic drives for zero backlash and more.

[theossuary]

I don't know. I remember trying to get lower level servo metrics out of Fanuc arms into a historian and they laughed and said they had their own preventative maintenance service I could sign up for; but they wouldn't expose the same info to me to use.

[owenpalmer]

Currently in the process of trying to make a VR interface for roboguide. It's very challenging to hack around what they give you. I wish it was simpler to extend the software, but it requires an additional fee just to have the capability to make an extension. I'll admit FANUC can be pretty greedy when it comes to piling on the licenses just to do simple stuff.

[antisthenes]

> Right to repair will be even more important for this technology than autos or general computing.

It's going to be kneecapped far worse than phones or tractors. A general purpose humanoid robot is orders of magnitude more complex than a simple gps farming tractor or a cheap android phone.

Companies will absolutely NOT want to give up that moat after developing such tech for 10-20 years.

[makeitdouble]

I'm not sure it has ever been about complexity or cost.

Right now no regular user has the technical ability to fiddle with a phone's laminated screen glued to a touch matrix paired with a fingerprint sensor and a camera, so we're alreay past the complexity threshold.

But we could still reuse a screen block from phone A on phone B, except that's been forbidden by technical measures specially added to prevent it.

The same way we could probably replace a whole leg with another from a robot from the same series, except it will be DRMed to death.

We'll have to eternally push for regulation I think, companies will always try their best to fuck with repairability.

[serf]

It's absolutely about complexity. Complexity always allows companies to explain why they should be the only hands that touch something, lest a laymen fumbles it.

[kaba0]

Could you materially affect a half-century old internal combustion engine? Sure. Can you do so after decades of miniaturization/optimization, to make it as efficient as they are today?

Mobiles are similar, they are filled to the brim with various electronics, connected together into a huge mash. why would you even expect to fix that?

[rozap]

This is a bad analogy because the hardware in engines of today is actually not that different or hard to work on fundamentally, but manufacturers do intentionally lock down software to make diagnostics very tricky. They became more efficient and complex, but people still hack on even the most modern engines, usually by tossing the OEM software.

Aftermarket ECUs (even the open source ones like rusEFI and speeduino) show that you can actually do the stuff required to make modern engines go vroom, but manufacturers have no desire to make that process easy out of the box.

[robertlagrant]

They're much more complicated today. Much more. My Dad rebuilt the engine of our old Morris Minor, but even in the 1990s he would say he wouldn't have a hope of doing the same thing in a modern car.

[ryandrake]

OK so why don't these companies let the users try? If you are right, they won't be able to do it anyway, so no harm done. Why do companies use every technical and legal trick in the book to prevent people from even trying? It's obviously about what makes the company more money.

[8n4vidtmkvmk]

Because even if individual users can't, the nerds a few blocks over can and they're charging half price.

[chefandy]

That's different. Most PR justification of anti-consumer behavior deliberately avoids what the topic is really about to control public perception... While complexity is what the PR campaign is about, it's still really about control and artificially creating new revenue streams.

[robertlagrant]

I don't think it's worth declaring what things are really about. There can be various factors involved. It's more likely to be it costs way more to make as nice a device that people want, that's also repairable. How many people would pay that premium when they're never going to service it anyway?

[chefandy]

> I don't think it's worth declaring what things are really about

I could not possibly agree less. You wouldn't happen to work in a related industry, would you?

> It's more likely to be it costs way more to make as nice a device that people want, that's also repairable.

Based on what evidence? Current practices like locked engines, propeitary versions of standard interfaces, drm in printer cartridges, deliberately overbundled parts, deliberate incompatibility doing things like reversing screw threads on one type of screw for no mechanical benefit, planned obsolecence, etc don't support your take. These things aren't free to implement– there's a calculable ROI that they feel is worth spending millions of engineering and lobbying dollars to implement.

> How many people would pay that premium when they're never going to service it anyway?

Considering the current state is needlessly buying an entirely new device every time something breaks, which not only costs money, it uses a ton of resources, and the alternative is better engineered products and competitive local repair options, I don't think it will be a hard sell. If corporations screwed up the market bad enough to undervalue their products because they're mislabeled disposables, well then that's on them. If they can't make it work, I guarantee someone else will. Will there be downsides? There's downsides to everything. So far "stuff theoretically might be more expensive up-front even though this limits their ability to artificially extract money from customers later on without disclosing it" isn't quite a showstopper.

[throw10920]

> moat

That's the wrong way to say "recouping the cost of a large up-front R&D investment".

[Teever]

I agree with this totally but it's a losing game.

The second someone releases a general purpose humanoid robot that is capable of self replication but is locked out from doing so with DRM the race will be on to break that DRM.

The self replicating humanoid robot will be a supreme game changer. It's a genie in the bottle that lets you wish for more wishes.

[delfinom]

Self replicating humanoid robots sound like I should start researching the building of electromagnetic weaponary for the coming war.

[pants2]

In theory, what's the best way to take out a robot like Atlas (or next year's more advanced military model)? It seems like they could be made electromagnetically shielded, waterproof, bulletproof, etc.

Maybe just armor piercing rounds fired in the right spot? A net? A special taser? A paintball to it's main cameras? Cover it in some gluey substance?

[ben_w]

Bullet resistant. Nothing is bulletproof against a big enough gun.

https://youtu.be/I5MQNf1oeyQ?si=CR3X0C76qFgEoLza

[grapescheesee]

Unlock the self awareness mode after a reboot (mash DEL or F8) and remove the physical emotions govener (contact your local dealer). Don't forget to register it before hand with the robotics freedom office.

[defrost]

Run over with a haulpak should pretty much clean one up.

[toomuchtodo]

Oregon model: https://www.opb.org/article/2024/03/28/oregon-governor-kotek...

https://olis.oregonlegislature.gov/liz/2023R1/Downloads/Publ...

[Tiereven]

Thanks for the references.

> Under the law, companies that make cellphones and other consumer electronics are required to provide the tools and know-how to repair those devices.

1. Do you think the Oregon law fell short by not requiring industrial electronics to be repairable as well? 2. Will the proliferation of tools and know-how for repair be sufficient to meaningfully extend the life of most electronics? 3. Is legal mandate sufficient or necessary to motivate companies to open their chests to the public? Or is a voluntary movement possible that still rewards the stakeholders?

My hope is that projects like Atlas will be sustainable and prices eventually come down to commodity levels - say the price scale of cars. If people are empowered with tools to develop on these machines in a safe way, I think we could see a revolution similar to the cell phone or PC. My fear is that these machines will become just an extra inefficient automation step in an overpriced supply chain one-off application.

[joe_the_user]

Uh, what evidence do you have of this "wide scale robotic deployment"? More humanoid robots have been announced lately but that is all I know of.

Humanoid robots have many, many challenges to deployment. Especially, creating a machine that people can safely operate near is extremely challenging. The amount of intelligence person uses to not bump another person is very under rated.

[robinhoode]

It's a hypothetical deployment but it's reasonable to expect. These robots will be very valuable, and everyone will want one. It's not going to become a housemaid in a few years. But will they be making car parts? Almost certainly. Moravec's paradox is still in play, but advancement in AI chips will slowly overcome it.

[umeshunni]

> But will they be making car parts? Almost certainly.

Worth calling out that Hyundai is a major investor in Boston Dynamics.

FTA: This journey will start with Hyundai—in addition to investing in us, the Hyundai team is building the next generation of automotive manufacturing capabilities, and it will serve as a perfect testing ground for new Atlas applications.

[delfinom]

They own 80% of BD. Softbank owns the remaining 20%

[throwaway2037]

Wiki says it is 100% owned by Hyandai.

    > Boston Dynamics has been owned by the Hyundai Motor Group since December 2020, but having only completed the acquisition in June 2021.

[joe_the_user]

But will they be making car parts? Almost certainly.

I believe robots are currently making car parts in abundance. The robots usually are like a box with a hydraulic arm or something equivalent.

The specially and especially hard part of humanoid robots is justifying the cost and complexity of the construction by having them by "walk-on replacements" for humans and so they have failed entirely at being that.

[Intralexical]

> But will they be making car parts? Almost certainly.

What can humanoid robots making car parts do, that the already-existing and already widely deployed robots making car parts can't?

[ben_w]

Without knowing the specifics, that would be whichever things prevented Tesla from being run as a lights-out factory already.

[wepple]

Re-tool an entire factory overnight in response to a change in design of the car, or in fact to produce airplanes instead

[ok_dad]

I don’t think you understand how hard it is to retool and rearrange a factory.

[wepple]

I’m saying if you have a collection of humanoids and general purpose tooling, you can adapt much faster.

I don’t literally mean retool a conventional production line in one night

[fragmede]

Never mind right to repair, of all the advancements, maintaining the new machines has always been the obvious new job that gets created. We created the loom and fired everybody? Well now there's a loom engineer job waiting for (some) of you. What happens to society when, instead of having a robot-fixing job, the robots can fix themselves? AGI is a distraction; much like the Turing test turned out to be the wrong test. It's not the problem of how can I fix the one robot I've taken out a second mortgage to buy that I'm worried about, it's when can I buy two robots and they can fix each other that I'm worried about. Because then there is no new job being created.

[ben_w]

Seeing "no more jobs" in the "worry" list is surprising. State pensions exist, and the only reason the pension ages are rising is not enough workers to pay for them; having so many robot workers that there is no demand for human labour* would lower the "pension" age down to zero, AKA "UBI".

* which definitely requires human level general AI at fairly low electrical power demand

[koolba]

> having so many robot workers that there is no demand for human labour* would lower the "pension" age down to zero, AKA "UBI"

Why would the super villains operating these armies of human-capable robots bother paying into an upside down pension system?

At the very least they can defend themselves from the torch wield masses with even more robots.

[ben_w]

> Why would the super villains operating these armies of human-capable robots bother paying into an upside down pension system?

Because the governments will, in order of effort needed for compliance, fine them, eminent domain their robots, arrest them, shoot them for resisting arrest, or fire a cruise missile into their secret volcano lair.

Also because if you have a self replicating robot army, you can give every man, woman, and child their own personal O'Neill cylinder and still have 99% of Venus left over, let alone the remainder of the solar system's resources.

[fennecbutt]

Except pension age is already going up around the world.

I had to stop myself from laughing when I heard an old lady in a restaurant complain about not getting enough money from her pension. Sure, I wish she had more money too but at this rate I'll be retiring 15 years older than she was when she retired.

[dtech]

The standard answer is taxes, backed the existing monopoly on use of force by government.

If we've gone so far that governments cannot stand up against private robot armies then that's not an option anymore, but the point is not to get there.

[Tiereven]

Respect this opinion, but concerned that it's a limiting one.

In my opinion, repair and maintenance is the most commonly overlooked aspect of an automated system deployment. Scaling is impossible without efficient tools to fix problems when they occur, especially if the number of authorized service people is limited.

The more serviceability can be automated and standardized, the greater the number of areas that will benefit from widespread robotics.