The main competition for robots are people and people are cheap, plentiful, and incredibly capable. It makes sense that robots would only be profitable in highly niche areas that were too dangerous for people or so repetitive that the bot is cheaper than the person.
It seems like we'd need something like off-world colonies, a fertility crisis or environmental collapse to really create a need. I think I'm happy with the status quo on this one.
That's true, and yet we already have a $20-30 billion (depending on who you ask) industrial robotics market.
There are a large number of highly profitible robotics companies in this area.
Fanuc with over $1 billion in net income from about $5 billion in revenue stands out[0], but also ABB ($2 billion in net income from $34 billion in revenue)[1], Omron, Kuka, etc.
They not be as interesting to report on as humanoid models, or the consumer market, but it's an enormous, rapidly growing market.
Yeah - those robotics company do exactly that stuff which is too dangerous or too repetitive for humans. There's also the issue of strength. Robots are a good candidate for handling materials when people literally can't. But, yeah, those robots don't look like people.
Here's an interesting fact. Most people think of forklifts as being stronger than people, and that we use them to lift stuff that people can't. That's true. But, what most people don't think about is why it can lift stuff like that. Hydraulic multiplication provides the power to actually do the lifting, but, a human shaped robot with the same hydraulics couldn't. Most forklifts weight about 3x the weight of a car and are mostly filled with lead. The same will go for robots. the form factor of improving human strength doesn't really benefit from changing the strength to weight ratio, it benefits from scaling both.
edit: as context, the average forklift is so heavy that when put on a trailer they exceed the tow rating of 1500-class pickup.
This is a mischaracterization of the size and operational use of the average forklift. Small (like under 5k curb weight) forklifts outnumber the big guys by an order of magnitude. If you count the small ones that the big box stores use I'd say it's two orders of magnitude. Most fork lifts are used to move small (<1000lb) things on pallets in indoor settings.
I personally think human form factor robots are a dead end but if the Incredible Hulkbot was as readily available as a forklift then then you wouldn't see so many things being palletized in the first place because you'd just have Hulkbot move them. The reason we have so many fork lifts is because we've settled on the pallet as a form factor and forklifts are good at moving pallets.
Maybe so - I looked around on practical machinist and on some of the website before making that statement and found that few items that really looked like forklifts were available that had load capacities under 3-5K, so those probably weight about 2x their load capacity. the kind of ride-behind pallet movers aren't really forklifts, but obviously do come with lighter duty availability. For reference, a pallet of bottled water weighs about 2000lbs. and most big box stores would rather operate forklifts with some capacity overhead.
I never thought of this and now that you told me it makes so much sense. Obviously a fork lift that isn’t heavy enough couldn’t lift heavy stuff because it would just tip over.
It doesn’t matter how strong you are if you don’t have the mass to balance what you’re holding. Mind blown.
> The same will go for robots. the form factor of improving human strength doesn't really benefit from changing the strength to weight ratio, it benefits from scaling both.
Yes, to lift 10t, you need the material and support to lift 10t. But a robot the size and weight of a child is much stronger and faster than any human and can easily kill one with the wrong movement. That's why they often are not allowed to work in the same location as humans.
There are currently exoskeletons developed to bring human strength back on par. But the speed race for anything repetitive is lost.
$20 - $30 bullion sounds much too small to be the robotics market. With all the fearmongering of “robots taking your jobs” I would have guessed the overall market was at least 10x that.
Just want to highlight the "incredibly capable" comment. While robots are excellent at some very specific things, they are absolutely horrible at some things an 8 year old can do easily. Folding clothes is probably my favorite example. While there are some "clothes folding" robots that have been released or are in development, almost all of them take about as much prep work as the time it takes to just fold clothes themselves.
Once we get to a state where a robot can just take a basket of clothes and fold (and sort) them like a human, I think you'll see an explosion of robot applications. However, I think that time is very far off.
I work at a company that provides industrial laser measurement systems using robots. This is a niche where robots are far superior to humans because they're much much more repeatable. Our clients are specifying their manufactured parts with .001" tolerances, and a human with a hard gauge is going to introduce a lot of noise in the QC process.
> It seems like we'd need something like off-world colonies, a fertility crisis or environmental collapse to really create a need.
Simpler than that: increase the price of labour such that robots doing menial tasks become profitable. A UBI would achieve this goal without catastrophic change.
A lot of the "cheap people" would be absolutely thrilled to be able to leave their "job" (indentured servitude is probably a more accurate description) in an asian sweatshop or an African subsistence farm and go to school. And no, they can't do that now because they can't afford it. Whether robots would materially change that is debatable. What is not debatable is that having "cheap people" is a better deal for the buyer than it is for the seller.
Well, it would give them time for starters. And it's possible that society would see the benefit of having an overall better educated population and decide to subsidize the cost using some of the newly generated wealth that robots would produce.
Or people to stop believing that you have to work to live. Imagine how the world would have to change if everyone decided to stop doing crappy jobs because they don't like them.
Not sure it really makes sense to fix a lack of people with something that's extremely labor intensive (at least up front). Not to mention, getting robots to be cheap requires a really highly advanced society, which requires a lot of people and a lot of time.
That said, it'd be interesting to see at what point or under what conditions a robot's total labor output exceeds it's total labor input (including watching it, debugging, programming, etc)
Isn't that because we haven't yet figured out how to actually get them to do complex stuff? If you could actually have a bot which could just go out and shovel snow perfectly, would you still prefer the status quo of doing it by yourself?
Mars rovers are incredibly complex because we had a reason to spend the money needed to build them. Who is going to pay people to figure out how to build a snow shoveling robot? It does not seem beyond our current level of tech, just beyond any current economic need.
You mean a snow blower crossed with a automatic lawn mower? Sounds like a quite simple and fun student project, but the hard part is probably making it safe enough for an actual product.
That might not be as true is the people were well paid, rested, happy and motivated. Robots are very easy to satisfy (some maintenance), if the human workforce was kept satisfied, robots would be cheaper.
I recently took a job with a robotics startup...it's a tough business to operate in with huge operating costs (compared to software), but I think there are a lot of opportunities yet untapped. The old Feynman saying "There's plenty of room at the bottom" is extremely true in robotics.
There are markets where robots have been (profitably) in service for a decade or more, but they are big robots with extremely limited roles: An arm that lifts a door onto a car, a glorified cart that follows a yellow line around a warehouse or hospital or whatever, etc.
The new generation of robots is human scale and has more flexibility built in. They deliver door-to-door/peer-to-peer, instead of along a fixed route. They carry small packages instead of a stack of boxes. They do inspections in places that aren't safe for humans. etc. Plenty of room for smaller, smarter, cheaper, robots.
But, it is definitely a tough/expensive business to build in, and the market probably doesn't have room for at least half of the companies currently running on investor money.
Once a technologically advanced product becomes truly useful and reliable, it's not a "robot" anymore. It's a dishwasher. It does one thing and does it well. General purpose robots are fundamentally too expensive to be mass-market useful!
Our goal as engineers should be to found the next dishwasher company, not the next robotics company.
> Our goal as engineers should be to found the next dishwasher company, not the next robotics company.
So I have thought about this a lot, and I even started and failed an ambitious robotics company, and I feel like a good niche product that could make a profitable company is self-cleaning litter boxes.
No, it won't be a multi-billion dollar company (at least not just with that product), but there is definitely a market for this and the current products are underwhelming. And technologically, it is about as approachable an unsolved robotics problem as you can find.
I know that one got canceled, but they're going to relaunch. Apparently a backer pointed out a design flaw (my girlfriend backed this project and has already been refunded her pledge but plans to back them again when they relaunch).
I agree with you for the short term. Advancing robot tech in general is better served by aiming to be the best X for a limited set of X.
A truly generic robot is probably several steps away and I think the correct niche for starting to approach that market appliance would be a human-scale (and moving) tele-operated human-shaped robot body. That would be useful for all of the places that humans are expected to navigate and particularly useful for situations involving danger.
I think we're decades into the era of that kind of "robot" and on the horizon there is a world with more general purpose robots. By that I mean, the mindless rote tasks are mostly already served by mechanical devices, like dishwashers, luggage carousels, various manufacturing "robots", etc. (this list could go on for days, if we've decided that all forms of automation is a robot). There isn't a hole for a "dishwasher" (where "dishwasher" is a placeholder for whatever simple rote job it fulfills) robot in the market today.
You have to look a little higher up the complexity scale, for roles that require low skill but high something else (so humans are hard to find to fill the role). e.g. say you need deliveries in a building, or even at the neighborhood or city-block level. It's repetitive, boring, work that is hard to hire for but can plausibly be done by a robot. It's hard to hire for because it is repetitive, boring, and requires a level of trust and reliability that a low-paying job doesn't really inspire in human workers. Delivery is a huge industry, when you consider all the scales at which delivery takes place (from tube systems and intraoffice mail systems on up to freight on trucks, trains, and boats), which could support dozens of designs of robot, and some of them are probably most usefully shaped like and acting like what we think of when we think "robot".
"Delivery robot" is still pretty special purpose, I guess...you could say it's a dishwasher, but it's a very smart, very expensive, dishwasher. It needs a lot of skills to do the job, and the closer it gets to the human it is delivering to, the more it probably should look like a friendly humanoid(ish) "robot" rather than like a self-driving truck (of varying sizes, based on role). A truck can't navigate elevators and doors, for example.
I like the idea of p2p delivery ("robot: take this saw I borrowed back over to my friend Joe"), but I think that's v2.0 of the package delivery bots which are in beta (or maybe alpha) right now, unless I'm missing out on something? Not to mention the legal implications of on-demand bots as drug mules or bomb delivery...
Sure, it's in the future, but it's not the distant future. The company I work with does P2P within a given environment. It has a map of the hotel/hospital where it works and can basically deliver anywhere on that map (barring stairs, and a few other limitations). It has no fixed routes, and is demand-driven. Someone calls it, drops in whatever needs to be delivered, and tells it where to go. It arrives, maybe calls the recipient to let them know it's at the door, and hands the delivery over, after verifying recipient identity.
There are many companies doing this kind of thing in various environments. Doing it at the scale of a city, like calling an Uber/Lyft, is probably a few years out, at least. Maybe in major cities taking the place of bike couriers it'll come sooner, I dunno.
"Not to mention the legal implications of on-demand bots as drug mules or bomb delivery..."
I think in the end, like most industries, robotics will end up being dominated by a handful of big firms. Each will have a mature hardware/software platform that is theoretically capable of building any a robot in any configuration you can imagine, but in practice the industry will settle on a handful of units with sizes and shapes that have proven themselves well suited to general purpose operation in a wide field of tasks. Most clients will standardise on a single vendor, and a particular unit and software pack, with limited customisation, and only Governments and very large companies will pay for completely bespoke robots.
Innovation and investment will shift away from companies trying to build a whole robot of their own to innovating on a single component type that fits within one or more of the large vendors supply chains, or plugging into standard units at the hardware and/or software level.
I think that's a reasonable assumption. Economies of scale apply to all hardware, and there are many roles that a given robot design could fulfill, given the right software.
The mainframe market might be a reasonable predictive model to look at (where there were a handful of dominant players, a few software companies around them, and hordes of service contractors who maintained them and helped companies accomplish their goals with the hardware available), at least until general purpose robots get cheap enough to look more like PCs. They're getting pretty close to that already, though, so maybe we'll skip right over the mainframe stage and right into personal robotics. I dunno. Depends how fast prices come down...I don't think the pricing works the same as PCs, as there is no "Moore's law" equivalent for robotics components. The price trends downward, and performance trends upward, but not at a multiplicative rate.
As it stands for now, though, robots doing complicated things are generally working as a team with computation in the cloud, so the PC "buy it and own it" model can't work. Robots, at least the kind that are doing real work, are purchased or leased as a hardware-plus-service bundle (again, kinda like mainframes).
Why is building robots so much more expensive than software ?
Software is complex. The main tools to create software cheaper are abstraction and reuse.
Can't those be available to new robotics startups ?
Is building a robot from standard components possible ? or using some sort of software based mechanical/robotic design tool, similar to the procedural modelling 3d graphic designers are starting to use ?
"Software is complex. The main tools to create software cheaper are abstraction and reuse."
Robots have all of the software complexity (more software than most software startups I've been involved in), plus hardware complexity.
Software doesn't have a supply chain to worry about. Software doesn't have to deal with an unknown environmental issue that makes the screen go black, or a vendor that started shipping you a different part while calling it the same part, or a sales pipeline that involves shipping around tens of thousands of dollars worth of hardware and staff to set it up and demo it, etc.
So, a software startup has to solve software problems. A robotics startup has to solve software problems on top of everything else.
"Is building a robot from standard components possible?"
Sure. Most robots in the market are mostly built from standard components with a few custom built bits for personality or to solve specific problems.
"or using some sort of software based mechanical/robotic design tool, similar to the procedural model 3d graphic designers are starting to use?"
Sure, there are tools for designing hardware. Robot companies use those tools.
But, it doesn't make building and shipping out multi-thousand dollar devices as easy/cheap as shipping software.
It's pretty straightforward – building a robot is like building a complex bit of software, except you have to build the complex software and a whole robot too :)
A robot platform can be a complex bit of hardware depending of the use case – it might be mobile, in an open/uncontrolled environment, using complex or expensive sensors, needs to be physically robust etc.
Now, many of those can be built from generic components – off-the-shelf controller boards, sensors etc. – but there's a lot of complexity in getting everything to reliably connect to and talk to everything else. Since there are far fewer robotics projects out there than software projects, there are also fewer of those "abstract components" in the first place – and physical requirements can mean it's harder to use them for particular designs.
Then there's the software layer. Again, depending on the use-case, this can be challenging - it can involve combining lots of sensor data, making decisions, planning motion, and so on. There are established algorithms and some reusable tools that can make this easier, but a lot of the magic is specific to individual projects. User interfaces can also be complex – building a reliable conversational UI is a project in itself, and if you're building a robot which interacts with humans, it needs to do this as well as all the robot stuff.
(FWIW I am working at a robotics startup right now, and every day is quite a challenge!)
Another big issue that complicates stuff is safety. In general, most software bugs don't have the ability to cause serious physical harm to a human. A bug in the robot software can cause physical harm (depending on the size and function of the robot, of course). Because of this, the software also is more expensive to write.
I don't work in robotics, but from my electrical engineering degree, I could hazard a guess:
- Software becomes much harder if you have to interface with hardware. You mess up software, you edit and then recompile/rerun. You mess up hardware, you need to buy a new copy of the hardware. Imagine if every time you wrote a bug or compiled a dependency with a bug, you need to pay $5 to somebody. If somebody else's software caused an error on a piece of critical hardware, your hardware may break too (e.g. car crash). And good hardware costs a ton of money. A Dynamixel servo costs something like $500, for a motor that's maybe 8 inches cubed or less, but if you use a cheap motor from China your precision over time will suck and software-based closed-feedback loops aren't as good as good hardware. So maybe you decide to do simulations, and you write a hardware emulator to load your software on. That costs a lot of money too, and doesn't and never will mimic hardware in all cases.
- Software for robotics is pretty low-level. You may need to write drivers for special hardware you develop, or you may need to rewrite drivers if proprietary drivers don't do the job for some reason. Drivers are if not hard to write, then a long slog to write (I don't write drivers). You need to make sure your software is highly performant, lightweight, and fault-tolerant. Maybe you need a real-time operating system, or you need to fork the Linux kernel and rip out bits and pieces (I heard somewhere SpaceX does this for their avionics, because they didn't want to deal with an RTOS). There is no let-it-crash philosophy. There may be no concurrency abstractions that are performant enough besides threading/locking primitives. There's the added complexity of failover that you plan into your design. You can't live without any of these most likely, because otherwise your customers may not live at all.
- By the time you master all the tools you need to build a successful robotics company, you haven't changed the system, the system has changed you. You accept the truth that good hardware is hard to make, good low-level software is hard to make, and great insurance plans are expensive. You're a Windows guy now because you use SolidWorks to design parts, and you can't be bothered to make a startup to rewrite SolidWorks for Linux because it's hard to get the physics/math background to reproduce even a part of SolidWork's functionality and user experience (which is a lot, CFD/FEA/etc). You deal with manufacturers who are perfectly happy with older tech because it works and because newer stuff always comes with risk. Your customers are enterprise Karens who shatter your roadmaps and don't care about your dreams.
This is a (possibly inaccurate) tasting of why building a hardware company is hard. I have the highest respect for those who do.
"Overselling is a dangerous strategy that can be counterproductive, even for the whole robotics community. Both companies and researchers publish videos of robots doing tasks, but sometimes they fail to point out the limitations of the technology or that those results were achieved in lab conditions. This makes it much more difficult to explain to non-roboticist industry executives the difference between creating a one-off demo and creating a real product that works reliably."
This makes me think of automobiles - cars were first seen as "a rich man's toy", but Henry Ford worked hard to change that perception. He said cars are for anyone. They can be used to go to work, to visit family, to go sightseeing.
I still don't see basic needs like that being addressed with robotics. For instance, is it going to take out the trash? Clean the toilets? Fix the plumbing?
The most numerous robot in the world is a vacuum cleaner. Autonomous cars are seeing huge investment.
Basic needs are gradually being addressed, but it’s true that you don’t see it. As soon as something robotic works, people forget about it, and this is how it should be.
> A machine learning researcher writes me in response to yesterday’s post, saying:
>> I still think GPT-2 is a brute-force statistical pattern matcher which blends up the internet and gives you back a slightly unappetizing slurry of it when asked.
> I resisted the urge to answer “Yeah, well, your mom is a brute-force statistical pattern matcher which blends up the internet and gives you back a slightly unappetizing slurry of it when asked.”
> But I think it would have been true.
The better we get at AI, the more we understand how AI works, and the more we can say, "that's not AI, that's just an algorithm that does <insert explanation of how the algorithm works>". OK, but there has to be some mechanism to how our brains work, right? How do we know we aren't just implementing shitty approximations of that mechanism?
Do you refer to roombas or the classic upright vacuum? Because it's hard to consider the latter anything other than a machine or appliance. It has one button, one function, two states, and is controlled by a human.
RoboCup@Home [0] is addressing such home tasks, eg. cleaning a toilet, in a competition setting for universities. This is an attempt by last year's winners: [1].
Taking out trash is a new assignment this year.
Note that in RoboCup@Home, the robot have to do a LOT of things well and its easy to be spread too thin over all those capabilities, whereas the Amazon Picking Challenge has a very narrow focus and you can still modify your environment a bit.
The robot. Would be interesting if the robot gets to a point where it's useful, reliable, and versatile for the home and its "owner". I wonder if this will spawn new economies as it did for automobiles?
Yeah, once we have a more general-purpose robot that can fold your clothes, set your table, cook breakfast, and load/unload your dishwasher, washing machine, and dryer (or at least handle a reasonable subset of those tasks) we might finally start to see some progress in this area.
Those are enormously complicated tasks to solve in the general case though, so I don't anticipate anything like that being available for quite some time.
> we might finally start to see some progress in this area.
You just ignored that the washing machine itself is a special purpose robot. Same with the dishwasher. Some progress already happened.
Go back a generation or two in the US, or live like the majority of people in the world, and you'd notice that laundry and dishes are very usefully semi-automated today.
They have feedback loops. Temperature, time and speed control.
It's an automated, feedback controlled, electro-mechanical labour saving home appliance. It doesn't matter if you call it a robot or not: progress has been made.
Even one of those tasks would probably be enough for people to consider adopting robots. People started to have telephones back when you could only do voice calls with them. Then, the tasks that mobile phones could do expanded, first to PDAs then to the iPhone in the smartphone age.
What I would pay for a clothes-folding robot. That is my least-liked chore, folding the same clothes in the same way every week. Yes I wear the same clothes every week - I'm an engineer.
> fold your clothes, set your table, cook breakfast, and load/unload your dishwasher, washing machine, and dryer
But are these pain points for a sufficient number of people? I can understand robots that could do these tasks would be a godsend for disabled people and the very elderly, but those are small markets. The vast majority of able bodied people just fold their own clothes and do their own laundry. It’s easy and takes minutes. I don’t think I’d even pay $100 for a robot that could fold clothes.
These startups employ people to do the laundry of their customers. If they get big enough, those startups might invest into automation.
Your dishwasher, washing machine, dryer, etc. is sitting around most of the day being idle, with its utilization being really low. Industrial laundry robots don't need to be mass produced in order to be profitable, they can be hand manufactured. This allows lower capital expenditures and allows for faster iteration. And you could pay some person to watch 5 robots and correct them when they do something wrong, something similar isn't possible outside of the industrial setting. So those services will greatly help with the invention of automatic laundry robots I think.
I feel like it was a disappointment that some of the brightest engineers of a generation went into ad tech, and next they're going into how to make the masses into part time servants with even fewer benefits than full time low-wage employees.
Most ad tech companies (Google, FB) likely devote well under 10% of their engineering might to work directly related to the advertising that forms their revenue stream.
Newspapers and magazines are ad-funded, too, but few would consider the work of journalists to be wasteful.
I would be curious to see an article, presentation, or quarterly report where that's written down because that's quite surprising. Maybe my gut is thinking ad-tech and all the support infrastructure, when the 10% is direct assigned engineering to ad-tech. And to be fair, the "support infrastructure" can and has other purposes.
Laundry is absolutely the biggest "housekeeping" item I have trouble with. I would pay a lot of money (thousands of dollars) for a robot that would take dirty laundry and wash, dry, and fold.
Yes, I suppose I could. I guess that I don't want another human coming into my house in my absence to do that, nor do I want to have to schedule it to happen when I'm home. Nor do I want to take my laundry someplace else and have to return later to pick it up.
I could probably live with leaving the laundry on the porch and having it returned there. I'll have to look into whether such a service is available in my area.
Traditionally, you would physically separate your living quarters from the areas where servants would work. In this scenario you could have a laundry room in the basement and a dumbwaiter connecting your basement to your closet, and have a separate door that only goes into the basement.
People will pay for convenience. These might not be particularly difficult or time consuming tasks, but they're also rather menial, and inconvenient to have to deal with.
Just look at how much money people will pay to avoid the inconvenience of having to physically plug in their phones to charge them, for example: https://www.amazon.com/s?k=QI+Charger
If you look at videos of factories in poor countries that make textiles, trinkets, and other low-margin goods it is extremely labor intensive because labor is so cheap. Eventually, as the world continues to be lifted out of poverty, robots will be cheaper than labor. But they aren't cost effective for many tasks right now.
There's another advantage for which human workers that don't get nearly enough much credit: flexibility.
When Tesla tried to create the world's most automated assembly line for the Model 3, they learned the hard way why Toyota and others have abandoned the idea: complex automated systems are, perhaps necessarily, tightly coupled and complicated. For Tesla, this led to cascading errors and insurmountable "debugging" challenges. Eventually it became evident to Tesla and Musk that, even at first-world prices, human workers were the better buy. [https://www.bloomberg.com/news/articles/2018-04-13/musk-tips...]
When you think about it, the work involved in building software systems may only be economically justifiable right now because computers can perform certain tasks many orders of magnitude more efficiently than humans can. In the physical realm, the differences shrink tremendously (or even reverse), making it far more difficult to compete with human workers.
On a trip to India I went to a Zoo where I took a bus to tour the enclosures for the tigers, lions, and bears. Between each species there were a set of gates to make sure the animals didn't mix. There were people who operated the gates rather than having an automated system that did the same thing.
To your point, in India it is cheaper to have a human being do that work than a machine, much less a robot. It will take a long time for the economics of robots to catch up on really cheap labor in the developing world.
Yes, exactly. Every robot from these companies I'd seen were underwhelming and nothing I would have spent money on. A robot that could do my laundry I'd pay $2000 for.
One data point: in NYC, wash and fold laundry service costs me ~$15 every 2 weeks. It would take 5 years (133 loads) for the robot to break-even and that's not counting soap or the time-value of money. Granted my clothes are relatively simple and non-delicate, but if the robot must handle delicates it will likely increase the costs as well.
The number one blind-spot for robotics projects is ignoring the availability of cheap (and often illegal) immigrant labor in most major locales. Western programmers think of most physical labor as abhorrent but working in a laundry in Western countries is often much better paid and comfortable than the alternatives back home. Plus your children typically have access to much higher education and opportunities.
A 5 year ROI isn't fatal. When people buy cars rather than taking taxis, the ROI is > 5 years. Many emerging technologies (solar panels, electric cars) start gaining market share when the ROI is in the 20-30 year range.
Solar panels are a $40B/yr industry! Many investors have done extremely well in it. The best time to invest is when the adoption rate is still low, not when most people already have one.
Yeah, that's been my observation too. I used to live in NYC and felt that. I empathize. Then I moved to Cambridge, UK, for some time and was surprised at how much dry cleaning would cost; it was quite expensive so I cut down on that.
With respect to labor, I also think robotics needs to consider humans as a variable in the equation, especially if they want to compete on a task that it's easy to substitute. This gets into politics, too, and it's a question worth thinking about. I wonder: as generations change, now that we're going from Baby Boomers to Millennials to Gen Z, how will this impact views and policies?
That's both a relatively cheap price tag for a robot and a famously hard problem in robotics. You're in luck though, foldimate is aiming to launch this year for $1k. Given that a washer/dryer combo can be purchased separately for ~$1k, all you'd need to do is load each machine. Ain't the future cool?
I looked at this because we've mostly given up folding clothes with our 3 kids. I watched a video of that machine and it seems to be more for people who hate the act of folding clothes instead of people who don't have the time to fold clothes.
In comparison, our Roomba is a time saver because it means we don't have to do anything at all to vacuum every day.
Yeah, the fact that you can't just dump a pile of cloths into the machine and get neatly folded stacks out the bottom makes it far less useful IMO. Still might save _some_ time, but not nearly as much as you might expect.
That's because it's fantastically difficult to manipulate soft, flexible fabrics. Building a [plausibly person-sized] robot that can reliably manage the entire laundry cycle end-to-end is hard to imagine doing on current tech no matter how much money you have. We're only going to see these sort of semi-partial solutions for a long time to come.
As would I, but I don't think anyone could make one that cheaply right now. I thinking of automatic milking machines for cows. Those things cost $180k each (which is why so few dairymen have them). I can't imaging a laundry machine for less than $20k right now...probably much worse.
I suspect that fixing the plumbing will require human-level intelligence. There’s a lot troubleshooting, strategizing, trial and error, trying to pinpoint problems in somebody else’s (undocumented) design etc. It’s basically like coding.
There are plenty of kitchen machines offering to save a lot of time(instant pot, foodie ninja, tovala oven and others), both at the home and at restaurants(see spyce).
They do their job pretty well. maybe not 100% but close enough.
Same with laundry, roombas, yard work, etc.
So now a robot gets all the simple jobs taken away from him, and is left with relatively little work to do. And this process will continue.
So maybe, in this environment, we'll never see a generic house robot?
Makes me think of electric cars. Just a few edits: "Electric cars were first seen as a Maker project or "a rich man's toy", but Elon Musk worked hard to change that perception. He said electric cars are for anyone. They can be used to go to work, to visit family, to go sightseeing."
Given that they're being built with autonomy in mind, wouldn't they qualify as robots owned by regular people?
basic needs are addressed by robots, they are just specialized ones called machines. I have robots in my house to wash and dry my clothes and my dishes! It is wonderful not to have to do these chores manually.
Robots are still too stupid to be used outside of niche applications such as military or heavy industry. In these applications, stupidity is a feature, not a bug, as a robot is expected to only do the tasks its owner wants it to do in the way its owner wants them to be done. But they still cannot be expected to work alongside people in their day-to-day because either they're not smart enough to interact safely with people, or people have expectations set by C-3PO and R2D2 that the robots just can't live up to (something the article touches on). Heck, people routinely get bored with Siri and Alexa because these technologies pretend to be smarter than they are, and ultimately outside of setting an alarm or playing a song, the value proposition they present does not even meet that of the stupider technologies people are used to. Real, exploitable stupidity is at the end of the day more valuable than fake smarts.
So if you want to be a roboticist and make money at it, don't expect to be Dr. Noonien Soong or Dr. Light. You will spend your career building dumb machines to perform mind-numbing tasks for the world's militaries, factories, or fast-food joints. Adding smarts to robots is a moonshot project, as Alphabet learned with Waymo -- you need to be rich or have a rich patron just to get anywhere. That's where things stand now -- maybe in a decade or two, things'll be different.
That said, I worked in robotics on such "dumb machines" for four years, and it was one of the coolest jobs I ever had. So if you can land a job in one of the niche industries where robotics is really useful, don't let my desexifying the industry dissuade you.
I am not sure I agree with the article's definition of robots.
The way I see it an automatic blinder that reacts on either a timer or sunlight is in the spectrum of a robots. It's a very simple one, but it does things for me that I would otherwise have to do.
Right now it seems like people are more trying to invest and push Robots in the image of "The Jetsons" which I don't think consumers are even close to being ready for yet, plus the technology and algorithms will always lead to disappointment until we get much much further.
We need to think simpler. The irobot is successful because it doesn't try to be a Jefferson robot despite it's name but rather it solves a specific need.
My kids get confused when I refer to my 3d printer as a robot. Heck, even the 2d paper printer is a robot. In that respect, I do have some very useful robots. I would very much like a robot that could do household chores though, like laundry.
It's frustrating. Even basic industrial robots have not been hugely successful outside of a few areas, such as automobile assembly. Amazon seems to have given up on the Amazon Picking Challenge; the last competition was in 2017. Here's the winner.[1] That's just bin-picking. Manipulation in unstructured environments is still very poor. DARPA's manipulation and humanoid robot programs got as far as public competitions and were then discontinued. Here's how far that got.[2]
On top of that, good robot arms are expensive. Startups announce low-cost ones, but many don't hold up in continuous use. Universal Robot from Denmark seems to be doing well, but their arms start around $35,000. Rethink Robotics went bust. Even with more "AI", the smartness is mostly limited to targeting objects that aren't perfectly aligned.
On top of that, good robot arms are expensive. Startups announce low-cost ones, but many don't hold up in continuous use. Universal Robot from Denmark seems to be doing well
Recently attending multiple industrial automation trade shows in China, I can tell you there has been a renaissance in robot arm production in the last five years. There are now many dozens of manufacturers deploying arms in real world manufacturing scenarios, where uptime matters. I can't speak for the quality, but if it's not there yet, it will be soon! One would assume this should effect some industry-wide price reduction.
>> Robots ... low-cost ones, but many don't hold up in continuous use. Universal Robot from Denmark seems to be doing well, but their arms start around $35,000.
$35,000 is the price of new, reliable car. and cars are much more complex, and suffer more forces trying to wear them out.
So why are robots so expensive ? is it mostly a matter of volume ? or is it something expensive about the tech ?
I think it is largely a matter of volume, Not just the volume of robots sold but also of the sensors and actuators that go into them. It's also a matter of market, no one is really selling robots in this price range to consumers, it's all going to large manufacturing companies or to large research institutes.
There are of course even lower cost robot arms, ~$100 toys, $1000 educational platforms etc. They just don't have the precision or power needed for industrial or research applications.
Would the costs and development be much lower and easier if the focus was augmenting humans strength in physical work? Would that still qualify as a robot company?
> Overselling is a dangerous strategy that can be counterproductive. Both companies and researchers publish videos of robots doing tasks, but sometimes they fail to point out the limitations of the technology or that those results were achieved in lab conditions.
This. Autonomous vehicles are the most obvious example of the overselling of robotics with the inevitable extending of the timelines that were promised a few years ago. The more we learn the less likely it seems we'll get AVs anytime soon, although anyone actually working in the area knew this already.
People were selling the idea of AV production systems that can solve hard robotics problems that havn't been solved in research in the most benign conditions. This was all driven by hype and money and it doesn't seem like there was much of a downside to those overpromising. Unless this changes and people are held accountable for their overselling then we'll keep seeing this behavior.
It has a genuine negative impact on anyone trying to do something that is actually possible and useful as their pitch will seem underwhelming compared to all the noise being generated by the people overpromising.
The short answer is that Robots are still heavily in the research phase, and not useful enough outside of some industrial zones currently to generate any profit.
That doesn't mean we shouldn't continue investing in them - they will most certainly become more useful and generate more profit. It just means at the moment, the best we can do commercially are more or less upgraded furbies and security guards who drown themselves.
Precisely. I think we're heading into an era where we need to have a "hunker down for the long haul" mentality, and yet we're heading into this at a time when our culture is particularly impatient.
The industry is up against several Hard Problems, and it may be a long while before any real gains are realized.
The near impossibility of building a robot to do something as simple as taking out the trash leads me to believe that the impending automation apocalypse is much further out than many people think.
There are limitless tasks that humans can perform with ease. There are very few tasks we've discovered that machines do with ease. Those few discoveries have changed the world.
In 2019, we can design and implement solutions for robots to do almost anything. But with very few exceptions these designs fall apart when they perform outside of a controlled environment or even inside of controlled environments when tasks need to be done reliably.
> The near impossibility of building a robot to do something as simple as taking out the trash
Part of this issue is that we think we need the robot to do it like a human would; in this example of "taking out the trash", we imagine the robot grabbing the bag of garbage from wherever, going outside, and throwing the garbage away, then returning.
Instead, what if the robot was the garbage bins themselves?
You have an inside robot garbage bin, and an outside one. The inside one would - once full - wheel itself outside and interface with the larger one to dump, then return. The outside one would periodically wheel itself to the curb to be dumped by the (probably someday fully automated - it's already most of the way there) garbage truck.
Essentially, make each robot a slightly smaller "garbage truck" that can be dumped by the larger units.
Of course, that still means the robot needs to be able to navigate a home, open and close doors, etc. If the home had automatic doors, that would make things easier. Then it would become a navigation problem mostly.
It's kind of like something I told a guy at my work who said he wanted to build a robot to bring him a beer. He was wondering about all the complexity, etc.
I told him - why don't you make the robot be the refrigerator, then you'd just summon the fridge to wherever you were, and it would dispense a beer, then return back to the kitchen.
Dish cleaning? What if the cabinets were dishwashers instead? Plumbing would be a nightmare - but in theory it would be possible.
Basically - think of ways to make the problems more tractable for the machine, instead of trying to make the machine act like a human (note the automation in many factories that does jobs humans use to do - they didn't make the machines/robots look like humans to do them, right? In many cases, the tasks and such were even changed for the sake of the machine handling/assembly - such home tasks could be looked at in the same manner, to an extent).
Thats the whole challenge we're dealing with now, right? Robots are trying to participate in a world where all of the interfaces are designed for humans, and thats why we get hung up on humanoid designs. I think its probably equal parts whatever the futurism equivalent of nostalgia is and and the raw practicality of making robots that understand human-world interfaces seeming easier than rebuilding the entire human world with robots in mind.
Keep in mind that automation is not just for manual labor. In many cases, it's easier to automate a white collar job these days with software than it is to automate a blue collar job requiring flexible physical tasks with robotics. To me, the big automation job killers are going to be autonomous vehicles, but then a slew of basic white collar jobs that people have just been doing for decades because that's the way they always did them.
with very few exceptions these designs fall apart when they perform outside of a controlled environment or even inside of controlled environments when tasks need to be done reliably.
Automobile Factories have gone to a model of humans minding robots doing a specific task. Early Roombas kind of fit into this model. They mostly worked by themselves, but they benefited greatly if you checked in on them to troubleshoot if need be. Broken washing machines and dishwashers are somewhat like this. There was one washing machine I had, where you had to wait until the end of the fill, then hit the top left rear of the machine like you were Fonz on "Happy Days," and it would magically work. Dishwashers work much, much better if people pay attention to how they place dishes, rinse dishes beforehand, and ensure the rotors aren't blocked.
DevOps fits into this model. Even programmers can feel like compiler/build process jockeys. (I often feel this way.)
I dunno the world where 99% of humans are employed taking out the trash that robots generate seems bleaker than the world where they're just unemployed.
Most hardware based consumer technology is derived from industrial technology. Let this be electrical components (motor and such), computers, dishwashers, or cars/automobile. Even though consumer purchasing power has been getting stronger these past two decades thanks to the large boom in software development, it's still not strong enough to push key technology. If the industry does not adapt "general" robotic technology, there is very little chance that consumers will see this in their households any time soon.
But the cellphone industry started in with the industry. Either from the early shoebox cellphone to the later model of Nokia and Sony Ericsson in the 90s. The main purchasing driver was for business use. It eventually got to the point of BlackBerry and iPhone which the consumer can buy.
When I saw Chobits as a kid I was excited about home robots, but I feel like Instead we got smartphones.
I still think home robots could be amazing, they could interface with you easily by voice and control everything connected to your network, act as GPS when they follow you to your car, go buy groceries for you.
But I'm well aware that, if a robot like that is made, the implications on privacy would be terrible (if privacy still exists by then)
> But I'm well aware that, if a robot like that is made, the implications on privacy would be terrible
I give it 5 years maximum before Amazon releases an "Alexa" robot that is initially just the same Alexa we all know, just mobile (flying or rolling, or both) with a small arm and able to do a few rudimentary tasks. There is no real loss of privacy to be expected over current Alexa anymore.
an "Alexa" robot that is initially just the same Alexa we all know, just mobile (flying or rolling, or both) with a small arm and able to do a few rudimentary tasks.
"Alexa, get me a beer!" Actually, a device or a couple of combined devices which could enable that gracefully might sell pretty well. (Some kind of fridge which can dispense canned beverages as the 2nd device? This could be adapted from vending machines.)
Even if they just tamper with immobile, non-fragile objects like physical light switches and door openers and if they follow you around as a mobile BT speaker/alexa, such robots would be useful.
Note that they aren't talking about Kubota, Cincinatti Milacron, etc. People that make robots for manufacturing are doing ok. But that also points to a big hole in this story, robots and what they are.
Back when I was running the Homebrew Robotics Club the question of "what is a robot" came up several times. Typically the definition is a machine that can perform some task while responding to changes in its environment that might otherwise prevent the task from being completed. The essence being sensing, executing an action to achieve a result, and identifying/signalling completion of the action. By this definition your dishwasher is a robot, it senses water temperature, dish placement, and sometimes even the amount of soiling on the dishes. It may adjust its cycle to insure the dishes are clean.
But most people mean mobile, anthropomorphic devices that interact with people when they think "robot." And making a business in that space is very hard indeed.
In part because we don't really have a way to "value" what the robot is doing. When you replace a union auto worker with a robot, you can directly compare the what you would have paid the worker in salary/benefits/training with what it costs to operate the robot. The math is "easy". But when you have a robot that is more "companionable" than not having a robot? How do you value that? And if you have tasks like "fetch me a beer from the fridge" that the robot can do, how do you evaluate the value of that versus doing it yourself?
Most people say "I'd love a robot that would pick up laundry off the floor and put it into the basket next to the washing machine." But if you say, "Here you go, for only $20,000 (ask about our payment plan), our picker-upper 1000 will collect dirty laundry and put it in the basket." people always blanch and say, "Heck for that much money I'll do it myself."
Which is a little weird, because a $10/hour job for 40 hours/week for 50 weeks/year is $20K. The robot can presumably work 168 hours/week for 52 weeks/year, which means that it's roughly 1/4 the cost of a human (and that doesn't count hiring time, HR training, sick days, worker's comp, taxes...)
The real issue is that we expect minimum wage workers to operate independently on minimal instructions, and as every programmer knows, explaining what you want done to a computer is much harder than telling a competent adult.
So what is the intersection of value and utility? Roomba meets the criteria, but nothing else? What about https://www.kickstarter.com/projects/tombot/tombot-affordabl... ? Or is that too light on the 'sensing' part to be considered a robot? Is a Furby a robot? The problem seems to be one of semantics.
They are working with a terrible definition of "robot". Most robot companies are profitable, if we are talking about all of the companies that participated in the huge surge of automation that swept of the Western world during the postwar boom of 1945 to 1973. Some of the biggest of the companies are still profitable. This is also a line of work that has, somewhat, consolidated into the hands of German firms. Siemens is probably the best known of the older companies, and SAP is the best known of the modern software companies, but there is a lot out there.
The more structured the task the sooner robots will be doing it. They've already eaten factory assembly where you do the same repetitive action every time. Now they're making their way into cleaning the floor, mowing the grass, cleaning up spills at stores, and (cough) putting items into boxes for online deliveries. Soon they'll probably be stocking shelves too. But a robot that can pick up your house the way a cleaning service can is probably the last thing we'll see before AGI.
EDIT: This is a bit risque, but I'm seriously asking: Has anyone applied Deep Learning to sex toys? The timing and processing of feedback involved seem like a good target for current tech.
People aren't generally fond of sharing their intimate encounters with a sex toy [1]. Some types might even need cameras, and that's definitely not flying.. How are you going to develop and improve the product if you can't collect the training data?
People aren't generally fond of sharing their intimate encounters with a sex toy...Some types might even need cameras...How are you going to develop and improve the product if you can't collect the training data?
However, specific people producing online media very much are. Many of these media people are effectively social media influencers, so marketing to them, specifically, might be a viable strategy providing access to a larger market. This is a population for whom making silicone impressions of their intimate body parts is just business as usual for producing merch. Products incorporating deep learning data based on their "performances" wouldn't be so big a step.
The indiegogo mentioned below for Autoblow A.I. Is mine. We published a paper on the machine learning study on bjs here: http://www.autoblow.com/bjpaper
I worked on a consumer robotics project which had great promise (in my biased take at least) but ultimately was canceled and scrapped due to poor management leadership in my opinion. It is expensive to bring a hardware group up and running, and I believe that there were expectations to create a class-leading product from nothing in under 5 years when it should have been at least 10 (see iRobot's Terra project timeline). It was a very exciting project and I am slowly getting over it's untimely and premature demise :(
The constant changing of requirements was a challenge but that is normal for any engineering project. I think the ultimate problem was the lack of understanding of the "Iron Triangle[1]" in project management. The company wanted it Fast Fast Fast Cheap and Good (in that order) but my team was trying to make it Good Cheap Fast (in that order). So, I think if we did it all over again we would have reprioritized from the beginning and dropped all the complicated design decisions we made, and work on those in version 2 just like any good MVP[2]!!
I think the key for a successful hardware project is strong technical leadership who understands the pitfalls and timelines of designing the mechanical, electrical, and software as well as can evaluate when and where to build vs buy across the project lifetime. With robotics in particular it is also important to understand how we can limit the environment to make things easier for the robot to operate, with things like magnetic or painted stripes on a floor, ultrawideband beacons, things like that. Making an autonomous system robust is difficult enough in a controlled environment!
[2] Some examples of things we could have completely dropped to save development time: Bluetooth and phone app, designing low cost custom motor drives, use standard materials instead of injection molded parts
I really wish product development failures were described in as much detail as successes. So much time is wasted by repeating mistakes someone, somewhere has already made.
1. Robots get classified as toys which sets an unreasonable price point.
2. Robots arn't good enough yet despite any hype you've heard.
3. In cases where a the product works better than the alternative and people are willing to pay a sustainable price the company does well ... just like every product ever.
Brain Corp in San Diego is solving this problem as we speak. In about two years, they've enabled floor care machines, a vacuum, and now an inventory tug for grocery stores.
Just checked out their tech - pretty interesting stuff. Leveraging other peoples machinery cuts out a huge amount of manpower needed to build a robot from the ground up.
Just from a logistics perspective hardware is harder than software to make profitable, and robots as a business involves a lot of hardware. More things have to go right, fewer things have to go wrong, etc. You rely on an entire supply chain to get things mostly right mostly on time, and that supply chain is mostly out of your control; you can have strong contracts, you can line up multiple suppliers for each part, you can vet them carefully, but at the end of the day you're tossing the dice that they'll deliver.
I've been working with a robotics startup for the past few months, and I'm often struck by how much extra stuff has to come together to deliver the product (in our case, we have to integrate with phone and elevator systems, have to deal with various regulatory compliance in every country we ship to, etc.). The sales pipeline is also long and expensive, involving shipping out multi-thousand dollar devices for evaluations, sending people out to demo the devices, etc. It's a production involving a half dozen people and thousands of dollars worth of resources to make every single sale.
So, yes, I think they are harder than some kinds of businesses. At least an order of magnitude harder than software. Though, because of it, the competitive landscape is somewhat smaller...robots are cool, so a lot of people want to work on them, so there's more than the market will support, but it's less pronounced than, say, the mobile app market in terms of competition for customers.
This makes sense to me. I think the word "robot" has some historical/cultural connotations that lead us down this rabbit hole of defining what a robot is. Rosie from the Jetson’s? R2D2? FANUC industrial robots? It’s tempting then to cherry pick one’s own definition of a robot, and point out challenges and theorize why it’s hard to build profitable robot companies. The question is fallacious.
In general, I think the hardware business is hard for all the reasons you mention. Or better stated, the hardware business has characteristics that make it difficult to get right the first time (more important in hardware than software, I’d argue) and scale. It’s not inconceivable for a single (or a small team) to build a useful software product that makes money (several successful mobile apps, companies like 37signals, etc.). It’s difficult for me to think of similar hardware analogs (I’d love examples!). This isn’t to say that scaling software—building companies like Google–isn’t incredibly hard. It is. But for a hardware company to even have a chance, it needs a minimum scale threshold from the very get-go (besides all the other things that any business needs – a useful product, luck, etc.).
I used to work for a large aerospace company in the early/mid 2000s, and the products we built were useful, and amazing (hunks of metal weighing tens or even hundreds of thousands of pounds flying in the air transporting people). I was absolutely awestruck by the fact that upcoming companies like Google and Facebook (which were “just websites” in my then-naive worldview) had fewer people, and made way more money than us.
Doesn't it all come down to: Coming up, and being able to make, something useful, that people actually want? Robots have an incredible potential to be helpful for us, but just looking at that social-media Jibo robot, I'm not really sure why I would need that.
Think about Vacuum robots on the other hand. It's still a small niche, but there's a lot of potential. I'd like to see some improvements in that direction, like if they were smarter, with better robotics & movement capabilities, they could be self-sustainable (charge themselves and empty their garbage bag etc.). Could do more tasks, remove dust, or check on things, turn off lights when you are gone... you can come up with some many things here.
It's simply too early. The product cycle is probably going to be very similar to smartphones. Two decades with lots of prototypes but negligible market impact (Apple Newton, PalmPilot, HTC P Series). Followed by a decade of rapid expansion when all the desired features can finally be implemented (like in the iPhone). We are probably still in the Newton decade.
There are a lot of profitable robot companies. They make washing machines for clothing and dishes. I have one to open my garage door. Like industrial robots they have people on either end feeding them.
I'm sure more robots will appear and some will be successful and called something else. (I'm a lot adopter of the vacuum, but there are others in developement)
“'I would say that a robot is a physically embodied artificially intelligent agent that can take actions that have effects on the physical world,' says roboticist Anca Dragan of UC Berkeley." [https://www.wired.com/story/what-is-a-robot/]
Industrial robots and Roombas may fit this definition, because they are controlled by fairly sophisticated software that synthesizes inputs from arrays of sensors to perform complex physical tasks. But less advanced machines like dishwashers and garage-door openers don't really reach that level.
Garage door openers do take input from an optical "obstruction" sensor, but the response is binary: don't close the door. In contrast, robotic vacuums translate distance sensor readings into a motion plan to cover every square foot of your room while avoiding obstacles.
That is a pretty narrow definition of robot. The term is often used to describe programmable, but not artically intelligent, machines. Like robots used in car manufacturing. I agree that washing machines shouldn’t qualify though.
I think robot implies complicated actions that are reprogramable to perform a wider array of functions.
Gall's Law : A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over with a working simple system.
It is very difficult to have a company culture that optimized both for software and hardware.
"Move fast and break things" doesn't work when you have to retool your production line or throw away inventory, or call a supplier to request changes.
Conversely, hardware development processes tend to be too rigid and slow when applied to software.
It is very, very hard to get these working together even (and particularly) when they are two distinct groups in a company.
I have had to deal with 'firmware' updates on a VM. You know, the full "build and download a binary" thing from some website, upload to FTP, then login to a website and point to that FTP server, so the binary would be applied and the instance restarted.
No big deal, right? We do that sort of thing for OS upgrades, right? Except it wasn't an OS upgrade or actual firmware, it was updating the application running on that machine. The sort of thing that could be accomplished by an automated CD job, and a docker pull (or heck, apt-get update). All because the thing was an appliance. So it was lifted and shifted as-is and now "it's a cloud app, because it's on AWS!".
I have had to argue against authentication for a HTTP health check. "But if there is no authentication, people can spam the endpoint, and it's slow and returns a lot of data". Ok, first thing: it's not a public accessible endpoint. Second, I'm not interested in testing auth or taking the machine down when credentials expired. And third, if it is slow, then just return a HTTP 200 for me if it is ok, I don't care about anything else. "But if I just return 200, it doesn't mean the app is actually working". Ok, so go do the checks at whatever interval you want, and return the last run. "But this needs authentication, otherwise there could be a DOS, as the checks are slow." No you don't understand, they are supposed to be async... This went on and on for months. I finally got the health check.
... but then, now they want to log all packets to our centralized logging, so that they can count them. Dude, expose a metrics endpoint, Prometheus will call it periodically. "But there are too many messages, we can't hold all of these!" No, you give me a counter. Return the counter. "But counting messages is even slower". No... you increment the counter _as you receive them, not when I request_.
sigh
So anyway. Even when hardware is amazing, software development can lag behind. Or vice-versa.
I see google as more of robotics company in making. They have software to identify things, surrounding and object like we humans do(which I think is a big feat).
They have a lot more pieces of complete robot problem.
Why wouldn't this work like personal computers? You use them at work or encounter them in school. When they're prevalent in those places, people begin bringing them home.
Hey, I could probably answer this one without needing to read the article, as I have literal TONS of robots in my solar facility.
You know those nice multi-tens-of-thousands-of-dollars Japanese robotic arms? We ditched them because it was just as simple to build our own robotic arms with basic framing and motors.
At a literal tenth of the price ($70K for our Kawasaki armature, $7K for our hand-built array of framing and motors with vacuum lines for picking up and placing solar cells.)
I guess a big problem is being realistic about the market needs. Maybe give up on dreams about humanoid helpers and focus on industries that actually are desperate for smart automation (e.g. logistics).
If you're interested in AI+robotics jobs check out https://nomagic.ai/ - we're hiring! Super exciting work with xooglers, academics and oxbridge/ETH grads.
It seems like we'd need something like off-world colonies, a fertility crisis or environmental collapse to really create a need. I think I'm happy with the status quo on this one.