Ex-Googler's startup comes out of stealth with simple, clever robot design
spectrum.ieee.org
spectrum.ieee.org
- go up stairs
- get up after a fall / reset itself
- pick up heavy objects
- precise motion (so no cooking)
- interact with buttons or touch screens (so laundry might be hard)
The one use case I think it might be worth it is if the robot can take a room full of toys and return everything back to where it belongs. Kind of like a one click, reset this room. If it can help me keep organized and clean around the house that would be a big win. Also, if it can help me detect which clothes are still wearable and which I should probably wash, that'd be nice.
I'd gladly pay 5k for a robot that can auto-organize my stuff around the house, even if it involves a lot of customization and setup. Imagine leaving your groceries next to the fridge and coming back to a perfectly organized fridge. Or letting your kids play in the play room and then the robot comes in and puts all the toys back. Or if you just tend to leave stuff out (throw your clothes on the floor or forget to return stuff), the robot can put it back.
I think that's a legit use case.
Actually putting things in their proper place is a chore, no matter how old you are.
You would probably want some bleach-safe clothing though.
Set an example by putting away your own groceries as soon as you put them in the kitchen. Boom 5k saved.
Maybe they will grow up to not be the type of person who leaves all of their crap on the table of the university table after they are done studying.
To be clear: I am not shifting on you as a parent. I am making fun of owning robots to do the specific things you mentioned. Obviously if you are immobile in some sense it would be great to have a cleaning robot
Open up jobs, pay a decent wage, and stop this line of thought about exploiting people as much as possible for as little money as possible.
Caregiving is a great job for those who don't have a job. More importantly, SV seems to be forgetting the human component of care giving. A robot _cannot_ possibly take the place of an actual human. It's just not the same.
Probably because low cost always seems to mean low income, low status, long hours, and insecure employment.
I own a house now with a dishwasher; I don't NEED it, but it sure is convenient.
Anyway, when it comes to kids picking up their shit, the important part there is that they learn the importance of not living in a fucking tip. If they or a robot or the well-paid help does it, it doesn't really matter in the end as long as it gets done.
I have seen recent college grads go over a year without washing/changing their bed sheets. In one case I found a couple dead bugs under the sheets. It may be pure laziness but in that case they are also feigning ignorance.
Obviously this evolutionary bias has gone a little off the rails with humans and we now have health and psychological problems as a result of efficiencies. I guess that's where self governance kicks in, unlike other organisms we can reason and adapt in spite of our fundamental evolutionary programming. Whether or not this evolutionary experiment works out is still yet to be determined.
I still say the same about dishwashers. In the time it takes me to sort things into what can and cannot safely go in the washer, scrape off the food residue, load the machine, and unload the machine, I can simply wash the dishes and stack them on a rack to dry.
Vacuum cleaners are useful but actually not very important in the kind of dwelling that the robot could inhabit because it will have to have flat floor with no carpets so a broom will do just as well (or a Roomba type cleaner).
If you have an ineffective dishwasher, you have to do lots of prep before putting dishes in, exclude lots of things from getting washed, etc. That takes longer, plus you don't cycle it as often, so you might find that you need a dirty dish and need to wash it by hand.
A very effective dishwasher is one for which you don't need to prep the dishes (just put them in) and can handle most of what you need to wash. It takes less work to use, and cycles more frequently. You can also cycle more frequently by getting a smaller machine.
The opportunity to invest in effective appliances is one reason it's nice to own a home, as opposed to renting.
Especially in the age of COVID-19 in the US, I can't imagine what it would be like to live in a super cheap crappy apartment like we were before.
Bonus, we also spend a lot less on electricity! We moved into a new-ish apartment building and I can keep the AC on 70 in Texas and my bill is less than $40/mo. In my old apartment, I would keep it at 76 during the day and only move it to 72 right before bed to save money. Bills were regularly $100+. Modern construction has so much energy savings built in to code now that you could potentially save money by moving to a more expensive apartment in a new building.
Most of the specific things they mentioned have nothing to do with being a parent, and also apply to people who do not have children.
That's the kids responsibility after they are done. Otherwise I do like the application to keep things tidy. It's a small thing but it improves quality of life and removes a common task of dealing with clutter. I think a lot of robots will start with huge limitations that nearly paint it into a corner. Eventually a more generalized platform will exist. Too many companies are attempting to make that today and skip the necessary steps in between.
Over here, in Eastern Europe i have no need for a car(i sold it this year, after 10 years).
Public transport is really decent if you use it outside of rush hours, most stores are in walking distance(not only form my location but all over the city). And there is a great network of cycle lanes so you can easily cycle to work.
Plus our cities are usually wide, but not tall, which means more places for developers to build houses.
- operate in your particular setting (it's likely over-trained for that specific environment)
- run for long periods of time as a fully autonomous robot
- be anywhere near small humans or animals
- do any of the tasks in the video with any kind of speed or precision
- learn new tasks
- handle multiple objectives, or even process the request of a single objective reliably
This is nowhere near being consumer grade. It'll either ship half-baked or you'll have to get on the "it's just around the corner" train, along with the autonomous vehicle guys.
It was to highlight that this really isn't the "do everything" robot that it's being advertised as that some are assuming it will be.
> The Roomba or any of the autonomous grass mowing robots
> can't do any of the things you mention either, and yet
> they are commercially successful in the consumer world.
No, but they are significantly cheaper single purpose robots that do their intended tasks well. This is aiming to be a more generic robot that does everything badly, at a much higher cost.
> there is a lot of potential here despite of the
> limitations
Well, the problem for researchers beforehand wasn't a lack of nice robot platforms to use. Some robotic arm controlled by a laptop on some wheels will already get you quite a lot of this functionality.
Really what a lot of researchers want is a nice framework that already offers a bunch of functionality they don't want to program themselves, so they can work on the more exciting cutting edge stuff. It needs to be robust enough to allow them to hack on random research hardware.
The Pepper robot falls slightly short on this for example, as many places have purchased one hoping it could be a robotic PoS/mobile FAQ - in reality it was just a mostly buggy experience for consumers.
I had a look at the code, it seems to run ROS, offer some possibilities, but I couldn't see at a quick glance some high level functionality they could hack on top of (although I could have missed it).
- You have to deal with and anticipate the behavior of dozens of other people/animals/machines.
- If you mess up you can injure or kill people.
This robot doesn't really have either of these issues. You'd have to be careful around babies and very small animals, that's true, but there's still a huge potential market even if you remove small pet owners and parents of toddlers.
Beyond that training the machine to identify common household items and tell it where to put them (the remote goes there, the backpack goes there, the shoes go there etc...) doesn't seem too far fetched. Being able to do the laundry (minus any ironing and assuming that you have a drying machine) also seems feasible to me. It won't be able to learn new tasks but if it's successful enough the company making them could continuously update it to implement new features, similar to how smartphone voice assistants keep improving and becoming smarter as engineers handle more corner cases.
Now are these features worth the current price tag, or even half of it? Absolutely not IMO, but that's obviously only for rich early adopters at the moment. If it gains enough traction I'm sure it can improve a lot and fast while cutting the price.
I could make a list a mile long of all the things my dishwasher cannot do... does that mean no HNs have dishwashers?
> infinite list of things it can't do?
When that infinite list is things people want. I speaking of course from the "consumer" perspective and outlining some reasons why this robot is many, many years away from being ready for that market. This particular robot will likely never be anything other than a research platform that has all support dropped for it after a few years.
> I could make a list a mile long of all the things my
> dishwasher cannot do... does that mean no HNs have
> dishwashers?
Your dishwasher is not an autonomous robot, arguably it's not even a robot. It does one very specific task mostly well. This robot is marketed as some generic home automation machine, which it isn't.
It only needs to be able to do juuust enough work to justify its cost (obviously not true yet).
Roomba is, for example, an even more limited device, but since it does one thing well enough to justify a $500-$900 price.
Talking about all the myriad of things it can't do is just pointless.
The problems I stated are really some of (but not all) the key problems that need to be solved in order for it to do as the advertising suggests. To put the problem into perspective, with billions of dollars man could send people to the moon, but is still unable to solve some of these fundamental robotics problems with similar funding.
I agree though, if this is ever to be a consumer product then it needs a massive reduction in scope and price drop (depending on what problem it solves). It's hard to think of a problem that is as commonly shared and simplistic in scope as hoovering without being a gimmick though.
You've got cooking which is basically automated through meal prep (e.g. microwave meals or even fast food services). You've got cleaning clothes - but the scope is enormous. Dusting and putting small items back in their correct location may be the only real possibility.
Record all conversations in your house
Record all movement in your house
Record every time you enter or leave your house
Record how long you spend doing tasks (watching TV, shitting, cooking breakfast)
Recommend products for you to purchase based on actions within your house
Recommend products to replace the previously recommended product you purchased
Send all of your recorded speech and movements to law enforcement
- Gather & sort laundry, maybe even put in the washing machine
- Gather dishes and maybe put them in the washer
All chores easy for an able bodies human. But imagine having some disability or getting older. Getting back that autonomy in your life is definitely worth the money. Heck, if I have to pick between playing with my newborn son or doing the above and the likes, I'd trade my car for a robot that can do this.
But oh no, having people around and paying them a decent wage smells like socialism! Anti-individualism! Better to pay 20K for a half-assed robot that can only do a fraction of what needs to be done than to pay someone's wage, right?
You could have a maid but then you need to have them around all the time or you need a lot of space. Also many places in Europe it's essentially illegal to have a maid.
Utter bollocks!
Presumably you aren't offering enough money. Anyway that is a far cry from illegal. Au pairs are not maids.
You'd be surprised at how much dexterity and precision all these things need. I doubt this robot can handle that.
The main equation for general-use household robots is: can it beat human fares?
A good weekly cleaner (4 hours) is about €22/h in Paris, or €4.5k/year.
Let’s say your robot lives 4 years, similar to your computer (which seems fair, since it is one, and an abused one at that). Since you amortize its price over 4 years, its initial cost must be below €18.3k ($21k) with human-quality-parity.
Personally, though, I would bet on dedicated, single-purpose robots, working together through wireless communication. Simpler to build, less costly, easier to replace. One small wheeled quadcopter for cleaning surfaces, one wheeled laundry box to carry it to the washing machine, one arm attached to the washing machine to put it in, etc.
The bigger issue, IMO, is avoiding distasters like Roombas trying to vacuum dog poop.
I'm happy with our Roomba. We don't have any dogs but is it really that bad!? :-)
Two months ago, my partner vacuumed up fox poop. No idea why he thought it was a good idea. We realised it wasn't when the soft, sticky debris coated the inside of the vacuum cleaner's flexible hose. :(
The smell only improved after I cleaned every internal crevice. Including submerging the crinkly part of the hose in a half-full bath and pulling through a sponge three or four times.
We still have to put deodorant beads inside the vacuum cleaner bag.
I agree, although I saw it as implied!
Robots are like AI: once they become widely used, they are no longer called robots. :)
To me, dish washers, mixers, vacuum cleaners are all robots that can be part of the household fleet.
Mixers and food processors are robots in French:
https://www.robot-patissier.org/robot-menager-robot-patissie...
https://translate.google.com/#view=home&op=translate&sl=en&t...
The thing about examples like the washing machine is, I haven't looked but I would be surprised if there isn't already a networked washing machine, less the powered door, that is meant for smart homes.
Robonix Philosophy at play here, inspired from Unix philosophy [1]
- Make each robot do one thing well
- Expect the output of every robot to become the input to another
- Design and build robots to be tried early, ideally within weeks
- Use robots in preference to unskilled help to lighten a chore task
Trying to stay as ethics neutral here.
We will probably see something like "Human assisted AI" with a sliding scale from mostly human to mostly AI for various tasks at various stages of development. The operators will be sourced where work is cheap, like Amazon Mechanical Turk.
Once VC gets it they'll fund the first few generations of bots on optimism alone.
Honestly who cares if it cant do everything. Nor can robotic vacuum cleaners and I have three.
@spykie So what if it cant go up stairs, have robots on each level. If it cant pick its self up, have multiple, let them help. Most things it cleans are light, so no problem there either. Cant interact buttons? Give it a rotating toolhead.
@bArray it doesn't need to operate autonomously, see the Human assisted AI argument. With good UX one remote operator will control five or six bots at once, tagging objects, issuing instructions. Give the User a good AR app, let them issue location based instruction around the house that the "AI" (remote operator) will slowly handle during the day.
Honestly, this space will be a primary design driver soon. I cant totally see IKEA making robot compatible cupboards.
If this space isn't huge I will be incredibly surprised.
Bezos & company aren't exactly considering the plights of their employees/independent contractors for their quarterly financial report.
Is it bad that I rank how successful a venture might be based on how many jobs it might destroy?
Mmm...I guess it's fair to assume that the IQ of a human cleaner is superior to anything resembling a Roomba. Would you hire a human cleaner and grant them access to your living area? And if so would you be monitoring their every move while they go about the job?
- 4-year robot lifetime is overly pessimistic. If upgrade is not the reason (which has nothing to do with lifetime, it's your choice to upgrade), a robot could last way more than 4-years (especially with right-to-repair laws in place).
- That said, such a robot will most likely be available for less than $21k and very soon in the near future.
I find interesting to address the middle one statistically.
A given machine has a number of random events that can kill it. Each possible event type i can be modeled as a Poisson process Xᵢ with, in a given year, probability λᵢ. The machine survives in a given year if no occurrence of event i happens, with probability Pr(Xᵢ = 0) = exp(-λᵢ).
For instance, a device with an average lifetime of 8 years and a single possible cause of death, has λ = -ln(½^⅛) = 0.087.
Given N independent event types (say, if you have on the order of N components), the yearly survival of the overall machine has probability Πᷡᵢ exp(-λᵢ).
Thus the lifetime of the overall machine is ln(½)÷ln(½^(Σᷡᵢ 1÷Lᵢ)) = 1 ÷ Σᷡᵢ (1÷Lᵢ) if each component has lifetime Lᵢ.
For instance, two components with an average lifetime of 8 years yield a single machine with an average lifetime of 4 years.
The more death-causing event types there are, the smaller the lifetime; and a general-purpose robot inherently has more.
I believe this result is counter-intuitive, and is the reason you might be led to think I am overly pessimistic.
> ... N independent event types ...
So if the robot blows a $0.03 (i.e., 3 cents) electrolytic cap [0], that's an independent event type that "killed" the machine? Just trying to get a sense of how you define "machine kill" (and as I said I mentioned a big caveat of right-to-repair laws).
For high risk people, this might be a good alternative for normally-simple tasks they can't do by themselves.
The knowledge workers automate jobs away, leading pundits to say the future for low-skilled workers is to provide boutique services for the upper class. The end being that the upper class is too ashamed to actually consume those services if it means looking at a person...well, that's sort of a delicious irony, isn't it?
I have also heard about people getting cleaners who initially do a good job, but whose work quality degrades over time. You wouldn't have that problem with a robot.
A safe assumtion is that all IoT devices spy on you. I wonder how much longer there actually are robot vacuum cleaners that are safely offline for sale with full functionality.
Hotels have a variety of tasks they would love to replace a host of low cost labor with robots with.
Someone in a higher income bracket wouldn’t think twice about that cost if it did significant work. On the other hand, the bar is high (cleaning, folding, picking up -- all are still fairly challenging AI problems, to say nothing of the "business" logic that would go into it).
Then you've got that quality servos are expensive. Some of that can resolved with volume, but not all of it.
In fact, France would probably be a very attractive country for this technology because of this.
The amount of time we used to spend making a single yard of cloth, or creating 10 pounds of flour is almost unfathomable today.
The prevailing economic theory was that we would automate away so much work that leisure time would prevail. Instead our massive increases in output have just driven greater consumption.
But the difference made by housecleaning robots isn’t the straw the breaks the camels back.
It would take a massive breakthrough in AI to eliminate the majority of knowledge workers at the scale that agriculture and textiles have already been transformed.
Lamplighters are especially useful to consider here.
How many lives has electricity saved? How many ICUs can operate continual machines because of it? How many people at risk of heat stroke can run AC in summer because of it? How many transplants and temperature controlled medicines got delivered because of it? What about its role in food safety? What about timely communications of life saving information, like those lost at sea asking for help, or severe weather alerts advising others to shelter? That's just the beginning, it seems incalculable really.
That lamplighter should have had a society willing to catch them during the transition. But imagine if it solved this problem by holding that one person's job sacred at the cost of millions of other lives instead.
It's absolutely good and relevant to worry about individual cases, but it seems like society is better off if we allow jobs to adjust to the actual current needs of humanity, rather than holding them fixed. Then we just also try to do our best to take care of as many of those who fall through the cracks as we can.
This reminds me of the little swarm of robots that come out of the walls to clean up spills in The Fifth Element.
It's a totally different use case to have a labor-saving device available 24/7. For example, I did not stop using my dishwasher when I hired a cleaner, and my cleaner does not do my dishes or put them away.
https://www.reviewed.com/dishwashers/best-right-now/the-best...
That's what I think the value is for this robot. A quick clean, reorg of the room / house.
This is where raising the minimum wage can spur innovation. If we gradually increase the cost of human labour (with higher wages, and other conditions), then the market will be motivated to fund & find technological solutions.
It's trying to work within the limitations of the robot and figure out what value it can bring that is the interesting part.
- stick a knife in your throat while you sleep
(Assuming it is always-on and connected to the internet, and has zero-day vulnerabilities which it obviously will)
Lots of washing machines and other household equipment now has APIs, so interacting with buttons or touch screens might not be necessary.
I think the 6x label on the dusting Spiderman scene is where I lost any interest. If it's that slow then it would take weeks to pickup the pile of Legos sitting on my floor right now, assuming it could.
Not being able to walk upstairs is a feature.
Cheaper option might be to have less stuff in the house :P
You don't need the robot for the first part. I never found the second part that valuable.
What makes you say that? All it needs is a "knuckle" on the backside of the grasping arm, to push things with.
We already make gloves that work with touchscreens, so the knuckle can work with them also. Vision is the harder part.
I mean I get that luxury and convenience of this robot taking care of stuff that you find boring or tedious, but imagine this robot being able to do certain normal every-day stuff that you're actually no longer able to do yourself? (Or maybe you're capable but it hurts to do, you get the idea)
A few years ago I had my leg in a cast, and I'm just thinking if this robot could've helped me put on my sock?
Kid's chores 2020: Up until 3AM programming the robot to take out the trash. Purchased the robot two weeks ago. Kid only moves to go to sleep. Trash is building up. No other chores are getting done. Chores dashboard is flashing red. Take out trash sprint way overdue.
In the distance: sirens.
The main chore I want a robot to do is laundry. I'd pay good money if a robot could simply put the clothes of the hamper in the laundry machine, transfer to the dryer, and then put them away.
Granted, you do have to take them out and put them away.
I live in an apartment with shared laundry
>Granted, you do have to take them out and put them away.
Which is by far the most time consuming part of doing laundry
how true
Originally seen from HN user idlewords; not sure if he coined it, though.
Edit: I have it backward:
Google acquired many robotics companies in 2013. I believe most of the acquired companies' leaders had contracts requiring them to stay at Google for 4 years.
Did Google delay the progress of robotic technology by 4 years?
The VC model obviously helps enable human progress in many situations. But surely not all situations. How can we know when the VC model will be a net loss for humanity?
I'm glad to see those folks moving on and following their dreams.
How many other areas are also effected by this 4 year lock-in period...
Sign here:
...and it's easily obvious that almost all of the new large tech companies emerged from VC.
Also it's an error of remarkable proportions to assume that "massive company" is synonymous with "net positive to humanity", particularly in tech.
Compare two cases:
1. Founders spend 6 months setting up a company and selling it to VCs. Then they develop for 2 years. The VCs want an early exit so they shop the company around and Google decides to buy. The founders and all employees must sign a 4-year contract or walk away with nothing. They sign. They waste 6 months on the acquisition process. Google mismanages the new teams. Google's CEO fails to put someone with customer focus in a position with power. There is no SVP of Robotics Products. Morale dives and then productivity dives. After 4 years, they all leave and spend a year recovering from burnout and setting up a new company to finally get back focusing on research. In 8 years, they have spent 2 years on high-productivity robotics research, 4 years on low-productivity research, and 2 years on logistics.
2. Founders spend 6 months setting up a company and securing grants. Then they develop for 7.5 years. In those 8 years, they have spent 7.5 years on high-productivity robotics research.
I believe there are other good ways to fund companies that have not yet been developed. VC does not fit all situations. Grants and bootstrapping have limitations. There are spaces for new funding models. What might they be?
The ideal acquisitions would least to synergy between the acquirer and acquiree. This would lead to a combination of reduced cost, improved products and the companies and consumers all win.
Unfortunately, it seems to me like this might describe a minority of real acquisitions, at least in tech. It seems like many acquisitions get fumbled and the original company and their products are now bad and literally nobody won from this but the owner of the original company having a nice excited. In many cases tough the goal of the acquisition even seems to be to eliminate competition which is terrible for the consumer and hurts the market. In that scenario I'd rather see companies "fight to the death" or find their respective niches.
I'm uncertain what impact it had on the market and industry that some companies now get started with the goal bring a acquisition by FAANG. I have no data or even anecdata, but intuitively it seems like something that would be interesting for an economist to do a study on.
The arm is not very rigid. Notice there's a barcode target on the end effector base (the "wrist") where the camera can see it. So the system can compensate for some bend. That's in some ways better than building a super-rigid structure.
I'm surprised there isn't a camera or two out at the end of the arm, for fine positioning. Cell phone camera parts are so cheap now that's a no-brainer.
Does this mean we get lots of hype and an inferior product that is eventually phased out an replaced by an acquisition built by non-google engineers?
The book I was reading that inspired it (which I thoroughly recommend) [2] talked about how telerobotics was going to help break down the last of three constraints on globalisation: moving people.
The other two are: cost of movement of goods (1st modern industrial revolution); and the cost of moving knowledge (the 2nd modern revolution, since 1990).
One of the things argued is that cleaners and dog walkers (for example) can't be moved, so a paradox of high skills economies is that certain low-skill jobs become more valuable.
$20K is getting frighteningly close to start making it more economical to have someone operate such a cleaning machine remotely from a low cost area than hiring a local cleaner. Skilled operators could even (with the help of AI?) operate dozens of these at once. Or they could do the 'easy' stuff, and get a 'proper' local cleaner in once in a while to do the tricky bits or a deep clean.
Interesting times.
[1] - https://zwischenzugs.com/2020/07/13/why-do-we-have-dev-rels-...
[2] - https://www.amazon.com/Great-Convergence-Information-Technol...
Also to take your example: walking a dog can also be considered a high-skill job. Sure you don't need a degree, but you need empathy and kindness. Pretty hard to teach to some people.
Buy a robot, and pay someone remotely to do your housekeeping for you. The housekeeper could even operate it from their smartphones.
The only downside is that you've got a real person in your house, but honestly if that's a hangup then you need to suck it up and just do the work yourself.
Slap on a VR system to the bot and I am guessing a lot of chores and stuff can be done remotely.
I would imagine a set of ready made extensions for various works, would be helpful as well.
(2) no "will the cleaner steal my golden watch / company's data" issue - you can always just replay the video and see exactly what the operator was doing / looking at
(3) ability to hire different people remotely - not just cleaners, but maybe repairmen, doctors, chefs, ...
If you can hire a remote operator to control a robot for half of the hourly wage that you would have to pay for someone to show up in-person to perform the same task that's still not anywhere close to economical if the robot also takes 10x as long, even if you ignore the up-front cost of the robot itself.
In order for something like that to make sense the robots would either have to become comparable to humans in the speed at which they are able perform their respective tasks, or the tasks would need to be at least partially automated so that an operator can control multiple robots at once.
- fall to low thousands price range
- do something useful (ie. not just a "social robot")
Motors, motor controllers + motion system are cheap enough for affordable home robots I think. The problem is automating actual tasks with the limitations of the motor controls.
They still can't do anything useful though.
And there are professional CNC machine costing waay more than this robot.
I think you are comparing apples with oranges. The price is also for a development kit. So I can imagine the price could be at least half if they sold a few thousand of them.
What's the use case besides novelty purposes? Who is supposed to buy this? Solely quadriplegic people come to mind but even that is a stretch and it's far too expensive. This is lightyears away from being a useful consumer object like a vacuum robot. If this is peak robotics, it is very disappointing.
No, it's not. $20k is peanuts compared to the cost of in-home care, especially since that would probably be pro-rated over ~5 years. If they can keep it in that price range and make it useful for simple household tasks (get something from the fridge/pantry/closet, put things in trash, operate a microwave or toaster oven, etc.), this could change a lot of people's lives for the better.
I don't care about teleoperation, for a home robot I want automation.
[0] https://www.goodreads.com/book/show/3201020-my-trip-to-alpha...
This has some potential.
$20k robot -> robot app/skill store -> Killer apps??? -> More volume ->... $3k robot -> More apps/skills -> $1k robot -> Everyone has one.
specifically:
1. what are the most expensive parts, and how fast is technology advancing (or costs declining) on those parts?
2. what scale is necessary for a comparable robot to retail for < $1000?
2) 5,000 - 50,000? Right now it's about twice as expensive per kilo as a Lamborghini Aventador, and they only made ~4,000 of those. It needs to be 18x cheaper to hit your price target, and that wouldn't be impossible at scale. A big chunk of the price tag is generous margin (because they don't know how many they'll sell) and paying for the software, which amortizes well. Those costs go away at scale, but then you need to spin up tech support and a repair department. I do tech support for educational robots, and I can tell you it's not... simple, or profitable.
Getting to that number would be hard. The design of the robot is wonderfully simple, (great! delightful!) but it's too easy to clone: the only complicated bit is the telescoping arm. Everything else is open source or commodity parts off the shelf. If they actually do sell a thousand of these, the Chinese will rush in and eat their margin. I hope either they have patents and some aggressive lawyers, or friends high up in the Biden administration, because if the design is successful, their company won't survive international competition.
From the way the telescoping arm moves, I'd bet dollars to doughnuts it's driven by a flex-tape linear actuator. Pretty cheap, lets you put the heavy motor right next to the mast, no precision bearings, backlash not so much a problem.
The whole package really is a great design, I'm a big fan. You can tell Edsinger has been using robots out in the real world, with regular people beating them up.
- Lidar on the base. RPi-Lidar costs a few hundred dollars.
- Motors. A bunch of Dynamixels according to the docs, also a couple hundred dollars each
- Carbon fiber tubes. No idea what those cost.
A motor may cost $5 but driver costs another $5 and cables costs another and you have to stack it against one another and if it has to support its own weights the motor must be better and if motor is better the driver etc etc and those keep snowballing.
Case in point: In China, floor cleaning robots (similar base motion and sensor requirements) cost ~1000RMB (~USD$142) retail. https://s.taobao.com/search?q=地+机器人&imgfile=&js=1&sort=sale-...
For centuries, real, physical actions in home life were improved with (mainly mechanical) devices subject to patent. The industrial revolution provides vast flows of consumer goods. Finance prospered (mainly) with loans, interest, and selling inventions like these, on large markets.
Fast forward 100+ years. Daily life is unimaginably easy, due to mass market goods for cheap, standardized living and generally, massive energy use. Markets constantly create products designed to be profitable given some game rules. Yet, there are diminishing returns. Vast, serious, ok yes urgent challenges face individuals and society, right now, and have been known for several decades.
An analogy might be the stunning success of mathematics in physics over the last 100+ years, sends multiple other soft sciences scouring math for certain curves or constants, because it worked for physics. Repeatedly, social science for example, is subject to physics-like curves and constants, searching for computable understanding, which is probably not appropriate to the endeavor.
The market is scouring product space, using formulas from long ago, to make the "killer" product. They say it here in the comments. This is to make money for people who are already very wealthy, using formulas that include massive diminishing returns (huge capital, you bet), and basically solve no real problem. Labor is over-abundant in modern times. People are living in isolation in Western Cities, and this is an isolation device. A very modest arrangement with a living person who needs work to earn money in the city, would solve the "problem" here, and requires no capital.
I suggest these talented, funded inventors look more deeply at the real ecosystem we are faced with overall, and get off of the gerbil wheel of get-rich personal consumption devices, because overall, societal survival actually does depend on tech right now.
You don’t really provide much of an argument for not using or developing a device like this other than there’s an abundance of human labour. What about if this device became much cheaper than human labour?
To your other point about other problems to be solved, if we only ever worked on the problems pivotal to humankind (e.g. global warming) then a lot of things that are useful but not pivotal to human existence would never have been invented.
I guess you don't have any disabled friends/family.
What if the dishwasher is sometimes run half-full?
And the arms can put clean things away!
It would be.... optimistic... to expect any household robot for anywhere near the same price to do that job as well and as durably as a conventional dishwasher. Which, I should point out, can wash dishes for years with no maintenance at all!
In fact, you can write news stories about the idea that dishwashers require any kind of maintenance? https://www.buzzfeed.com/kristatorres/clean-dishwasher-filte...
I am not ordinarily a pessimistic man, but I would expect to see major, society-destroying levels of technological unemployment from advanced automation long before we see a dishwashing robot.
Let’s be honest, the roomba sucks, so It’s not really a counter argument.
Once there is a robot that is useful and also seen as a status symbol, that’s when the feedback loop will begin.
That was buried way too low in the article for most HN commenters to bother getting to. Maybe this fact will help refocus the conversation from what this thing can't do to what it might be able to with some more work. Personally, I think it looks innovative. Adding a relatively inexpensive telescoping arm and some LiDAR to a Roomba could be another evolutionary step forward in home robotics.
I can easily imagine wiring this thing up to Alexa and asking it to go get me a drink from the fridge. That alone would be worth a few hundred bucks when it becomes a commercial product.
Stretch™ Research Edition
A new kind of robot.
[…]
A Robot for Researchers by Researchers
Any robot that can move as fast as a human and perform similar work to a human will need to be close to as strong/heavy as a human. And humans can kill each other with their bare hands; we just know not to. A Kuka LBR iiwa can only lift 7-14kg, but could easily accidentally kill/maim someone with common payloads of that mass.
Obviously there's also some slowness w.r.t. the actual control here (i.e. safety pauses, planning latency, etc.), but it's a tall order to build any kind of home robot - especially one that has such a large workspace - that is intelligent enough to be both fast and safe.
It will need bigger wheels or legs then because it would be stuck in one room in my house because of the thresholds at each doorway and the rugs and carpets.
And my house has three stories so I would need three of them because it can't climb stairs.
It's getting tiresome to read crap like this because while it might work in a dwelling purpose built for it it is guaranteed useless in most dwellings occupied by real people right now.
I remember a reacher being a pretty successful present last time I played "white elephant gift exchange" :-).
1: https://www.amazon.com/Reacher-Foldable-Lightweight-Reaching...
There's a few "grabbers"/"reachers" that have a gripper that looks pretty similar to the Stretch's; I'm not sure if there's an "original" that the others are clones of: https://amzn.com/B07RR6N9C7 https://amzn.com/B0812B4TF1 https://amzn.com/B01MAW25UW
There's not much to the robot besides the arm, so, um, yeah?
https://scoutsomerville.com/wp-content/uploads/MG_4809-1.jpg
https://boingboing.net/2016/06/08/the-amazing-shitty-robots-...
I imagine a bit of the current price tag is "we want serious purchasers who will be partners in the development of this device"; they don't want to build them for people who will just dick around with it for a bit and then relegate it to the corner of the room.
What I'd like to see is standardized interchangeable parts and configurable software so we can spec robots to their tasks and benefit from reuse of parts and code as much as possible.
On the other hand, those Boston Dynamics robo-dogs with their snakelike extensions reminds me far too much of the Fido enemy (a.k.a. Parasite) from Quake 2: http://vignette1.wikia.nocookie.net/quake/images/d/d2/Parasi...
EDIT: And I should add, I hope they market this to that severely underserved market. If they only market it as a product for the rich to do simple tasks, it will suffer Segway's fate. If they create a market for it in healthcare and partner with those already in that space, this could be something big. It would pay for itself for people who need 24/7 care, if it were reliable enough.
how would you automate that? by building robots to carry the water containers? what is the optimal solution may not be the most obvious.
A bit more seriously, this is a research platform. Maybe in 5-10 years it'll prove to be the seed of some interesting tech, but right now it's just a lab experiment. If you buy this and expect it to do basic housework for you, you are destined to be disappointed.
Most normal people would take that as a sign that the founder has already had a sucessful career in tech, because "FAANG" don't hire mediocre workers. The journalists knowing this trend, use this information to communicate this to people, speaking in their language. (Also this language makes some sense as well.) I don't see a problem.
Their propaganda is working well I see. Having a Stanford degree and being able to answer CS questions (two things FAANG companies love) does not guarantee you'll be a good worker.
Having worked with robotics in mfg, part of the cost is stiffness. A stiff arm and mounts means repeatable actions.
In the video this thing was flexing all over the place. It’s cool, no doubt. But it seems a well placed LEGO could nearly shut it down.
I’m not seeing 18k as “cheap” in any sense here. Maybe I just don’t understand the business model or what goes into this.
I don't know anything about home robots, but to me this seems to be the first design that is practical to bring to consumers. Not 2 human arms (or even legs) and some expensive face, but instead just a Roomba-ish base, telescope arm with grabber that moves up and down, camera on top.
So I'd guess that with sufficient volume this could get down to the $5k mark.
That said, I doubt this bot will ever make it to the stage of being a consumer bot. The sorts of problems that need to be solved are on the scale of that of autonomous vehicles, probably even harder.
Universal Robots are popular for research, and from what I heard start at around $30k.
KUKA LBR iiwa is more like $100-200k.
Let's not pretend a consumer model is even remotely ready though, these scenarios are highly contrived. The goal at the current time is research, and at that price point there are a few options available.