The killer product of robotics could be a low-cost collaborative manipulator
lorenzopieri.com
lorenzopieri.com
I think the killer app - and something I would pay for immediately - is a moveable platform programmed to travel between waypoints.
So, a platform that stays level regardless of terrain and you put stuff on it and tell it to go to the (living room, backyard, carport, barn, whatever).
I arrive home with a load of groceries and I put four bags on this platform and press the "kitchen" button.
That's it. Just a magic box that transports (slowly) things to places around your yard/property/warehouse.
it can clean the plates on its way back
Checks username
Ahaha
Then, you can drop it at a second place you shine the laser.
(The robot itself could be a kind of dumb RC car kind of thing that you are basically driving by shining the laser)
A low speed, low accuracy robot was tried with Rethink Robotics, founded by the Roomba guy (Rodney Brooks), who tried to compete with Universal Robots and got their clock cleaned. Nobody in industry wanted a slow, inaccurate robot, even at a good price.
What's holding us back in robotics is not the cost of hardware. There are plenty of use cases of $40k robots that are still not pursued because engineering special end effectors and integration is too difficult, not because of the cost of the robot. Reducing the cost of the robot to $10k makes no difference in those applications because it's not the gating factor.
The idea that companies will just figure out folding clothes and other complicated tasks because the cost of hardware falls is, I believe, fanciful, but I hope I'm wrong. Maybe "AI" will hand-wave solve it.
Also, if someone was throwing that many resources at a given problem and it was to sell pretty cheap robots, how would they get their money back? Lots of expensive tools sport expensive software.
My kitchen is full of cheap, perfect robots. A microwave with cook sensor, a fridge that makes ice, a coffee pot with timer and brew settings, a pressure cooker that somehow cooks everything perfectly all the time.
In my garage I have a door opener, a clothes washer, a clothes dryer, some motion sensing lights.
In my backyard I have a sprinkler set that waters the lawn.
On and on ...
Those are the killer robotic apps, and we've been refining them for 100 years. When I teach classes, I honestly get a laugh when I say that the toaster is the most perfect robotic design of all time from a customer service perspective.
Simplicity (at least simplicity of functionality and presentation) is best if you want ubiquitious markets, I think.
Most of those execute a fixed program with no feedback. The hallmark of robotics, as I understand it, is complex feedback mechanisms. Robotics classes at my school were 80% control theory.
As things become more simple and robust, I would argue the degree of adaptivity will necessarily decrease, because adaptivity is required to handle situations off nominal, and those can and should be designed around.
But actually, a food processor. It isn't 'quite without direct control or supervision', but the direct control is more like loading feed-stock and the supervision level is similar to using an oven: you're mostly there to make sure things don't go incredibly awry. I used to think they were kind of goofy in a home kitchen and had only ever used one at work, but after developing RSI in both wrists and working on dinner at a friends house I started to see the value. Want to make fancy mac n cheese? Pop that block of the good stuff in, shredded fast. Same deal for hashbrowns. Or really any vegetable you need to have some coarse/fine chop applied to.
The home is always going to be a nightmare environment for this kind of thing, though— even without the mundane issues of stairs, doorknobs, appliances with varying interfaces, and hazards like pets and clutter, there's just far too much breadth of potential tasks to want it to do.
Later companies understood that zeroing in on a few tasks to do well and repeatedly (delivery, laundry, floor-cleaning, etc), in ADA-compliant environments (hotels, hospitals, airports), was a much less scary place to start.
We have two choices: 1) Design robots for places optimized for humans, or 2) Design places optimized for robots -- a future where the living space (and all the machines and other things in them) are redesigned from scratch for the robot that will cook, clean, and move around there. Then a second priority is making it comfortable enough for the humans to also live there, too.
I suspect it's easier to adapt a human to an environment optimized for robots than it is to adapt a robot to an environment optimized for humans.
Automated kitchen - place a bunch of ingredients regularly or right before ordering a meal, pick up freshly made food of your choice (or what's possible).
Automated laundry - dump dirty clothes, pick them up washed, dried and ironed (probably the hardest part heh).
Automated garage with charging/fill up, maintenance checks, washing?
Will need a lot more space, so it would be only for the rich, at least at first. A common automated kitchen for an apartment complex might be possible.
And better make them very reliable or maintenance can be a nightmare.
I am dubious about the last line regarding maintenance making it into reality though.
Given current track records on "planned obsolescence", "right to repair" and McDonalds issues with their frozen machines I suspect such maintenance will be the new "subscription".
Replacing existing human spaces with robot-optimized spaces would make robots significantly more expensive than basically any other approach.
But once you begin describing that, it's clear how absurd it would be to install it all for use a few times a week.
Home mobile manipulation will never be a thing until it can use the existing HMIs.
Houses built over the past 50 or so years are designed to disintegrate over 50 or so years and be rebuilt anyway. Although demand for new construction is reduced by the way that neighborhoods tend to go from high income to low income as the buildings physically disintegrate, there is a lower limit somewhere, and redevelopment will happen everywhere eventually.
This may happen in places, but in any case building larger (to accommodate robots) spaces is not a way to make the robots more cost-effective or drive adoption.
My biggest problem is that if that world comes out of the current one, it's going to be one heavily mediated by capitalism (just like everything on my iPhone). So, yes, the robot could scrub the dishes. But once I've accepted that relationship with my robot, version 2 comes out and it has a feature where it doesn't need to spend 8 hours doing dishes because it includes a free trail of the $20/month dish subscription and just throws them out and signs for the shipment of new ones that comes while I'm away at work.
Whatever the exact scenario, the environmental cost will be huge. The COST cost will be huge. And probably the robot will be reporting the details of what I ate and what stayed on my plate back to the company to sell on to the ad sales branch of the company
Life insurance companies and your employer.
The problem is they're starting this effort from the basic joint being an electric motor. When you start there, this all follows as what you have to do to build a robot, you're forced into a corner in your design. Instead, it's far better to have the safety and operation come inherently from the way the product is designed.
I saw an impressive demo at an automation conference about a decade ago, it was a cobot where the motion was based on inflation of pneumatic bladders. It functioned similar to the way biological muscles work. It did not need gear reduction, or force sensors, or brakes, or high-speed amplifiers. If it ran into something (or something ran into it), it didn't strip gearboxes, the bladder just complied.
I think that will be the 'killer cobot' someday, but it's going to be hard to impossible to launch into the industrial space.
I explain: we already have an incredible successful robot. It’s the roomba (and it’s imitators). It’s a robot, but suddenly it’s not the robot we’ve all been waiting for.
Just as Siri and Alexa and Google Assistant suddenly are not widely seen as AÍ because, even when they can talk, recognize your speak, search for content and do some tasks, they are not the artificial intelligence we have been waiting for.
It seems we’ll always raise the bar, no matter what.
200 years from now, taking, walking, autonomous smiling android: not a robot.
I always assumed it was going to be teledildonics.
..damn, a downvote. I mean, who can be against teledildonics? That'll be the main thrust of technology when our robot lords and masters take charge.
I was more convinced by the author's pitch than I thought I'd be, but the problem I see is that the technology just isn't quite there yet. Thus it'll probably come out to be more like the early PDA's than the iphone.
Still, the general direction could have more promise than my first judgement of it based on the title.
They can reach into the car's model of the world to help it understand e.g. this thing that you decided is an ambulance is actually not an ambulance, so you don't need to treat it like one. Or, very simply, that road is shut, don't try to use it.
Even worse, I have a smartphone which can do everything a smart speaker does and more.
The author underestimates the physical requirements for most tasks we do day to day. 0.1mm precision is roughly the accuracy you need to put a #8 screw into a clearance hole toleranced for a loose fit. One might think when loading a dishwasher that you don't need to be precise because the dish can go "anywhere", but in reality that's just because a human fundamentally understands where plates can and can't go, and can adjust their movements accordingly. If you are trying to place a dish on a granite countertop, and the plate goes 0.1mm into the countertop, you're going to notice. Such a robot arm would just barely be able to peel a potato.
5kg of payload may sound like a lot, but it really isn't. For starters, that doesn't include the weight of the end effector. Then this is the max load that the robot can support, but if you need to manipulate something quickly and precisely, that further reduces max payload. Holding a single item of clothing should still be fine, but what about say pulling on an item of clothing out from under a pile of laundry? Or how about lifting a pot of water off a burner on your stove? Pouring a glass of milk from a gallon jug would be close to the limit of what such an arm can handle.
I'm not sure where this figure comes from. In inflation-adjusted dollars, the Apple ][ cost ~ $6,000.
> Previously shown off to a few thousand rabid fans at the West Coast Computer Faire, the Apple II’s arrival means the masses can finally get their hands on the breakthrough machine. A base unit costs $1,298 — the equivalent of $5,810 in 2021 money.
https://www.cultofmac.com/484382/tiah-apple-ii-goes-on-sale/
Maybe all it takes is a good, but overpriced robot selling for around $6,000. The high cost of Apple's first computers was a major driver for a work-alike computer market that included Atari, Commodore, TI, Coleco, and Tandy.
If the Marketplace Disruption view of PC history is any guide, I suspect the first wildly successful robotic product will cause professionals to turn up their noses in disgust at the triviality of it all. They'll bemoan the lack of practical applications. They'll cluck at all of the things it can't do. Meanwhile, a small group of buyers will find the toy-like things irresistible. This product might have already been introduced.
Also, it's instructive to consider all of the things that were predicted for home computers vs. how things ended up turning out. Many of the concepts were sound (games, recipes), but few thought of the role a ubiquitous, worldwide, fast network infrastructure would play in creating entirely new classes of uses for small computers. And almost nobody predicted the astonishing rate of miniaturization/performance increases that would occur and which continues to this day.
The "iPhone of Robotics" probably looks more like personal ownership of several robots: Roomba, iLaundry, iFoodPrep, iPutAwayGroceries, iPutAwayDishes, etc. iRobot has built a solid business reasonably automating a single household chore. I would bet an all-in-one laundry machine or bathroom cleaning robot could also do compelling sales.
At a minimum, it seems very ambitious to try to attempt to solve via robotics problems that are not well-solved using single-purpose machines.
The toaster is a perfect robot. I mean perfect. A more modern example is my smart pressure cooker that does just about everything in closed-loop control. It's fantastic.
Automation is not only a very old practice, it's actually stagnating relative to the amount of magic miracle appliances that appeared in the first half of the 1900s.
You have a faulty model, that's not a knock on the toaster as a robot.
Why do I have to put the bread in the toaster? A toaster should load itself.
And half the time I can't get the toast out without using a knife. Why doesn't the bread come up to a level where I can pull it out without burning myself?
Why can't it ever toast the right amount? It's always too much or too little. It should be able to determine everything it needs to about the bread to toast at the correct level, no trial and error and no wasted slices.
In fact, I disagree with your statement so much that I'd ask the question, why are toasters so shitty?
Probably a lot of the answer is cost cutting.
By the way there are already self cleaning cat litter trays and they’re absolutely awful and prone to poop-related bugs. I’m not sure any robot could reliably react to the developing situation of having a wet turd on the litter scooper.
For example, you might have a little robot that can have an attached rotary gripper, for ensuring that the stove top is off. It sits on the stove knob and can turn it based on some input. Or one that can open a deadbolt from the inside. Or one that can brace against a door frame and turn a door handle. You could even include a tiny, low resolution camera that can snapshot the current state of the thing that the robot is controlling.
Having these devices be adaptable, with different attachments, would be important to making sure they work in the most situations. Whatever attachments you put on, and how you position them, are unique to whatever analog device you are interested in controlling. Some could brace themselves against the surroundings, some could be screwed into a wall. The key is options.
It should be able to cook good tasting homemade meals from real ingredients (even fresh from the garden if you have a garden), not just plastic wrapped junk food. You load food items through hopper or door in one side, then a manipulator arm picks them up, reads or otherwise identifies what they are, and stores them in the inbuilt pantry or refrigerator.
It should keep an inventory of the refrigerator and panty and be able to suggest recipes that use what you have on hand while optimizing for cost, nutrition, and minimizing wasted food from spoilage. It should be able to produce shopping lists for you based on foods you use frequently or recipes you want to cook in the future. (It could even auto order foods and have them delivered, although I like the idea of just getting a shopping list and buying it myself better.)
It should be able to help you avoid wasting leftover foods by reaccepting leftovers, storing them, displaying the leftovers as a suggestion for the next meals so you don't forget, and know the optimal way to reheat or re-prepare them.
It should be controlled through an app on your phone, over the local network, and definitely not dependent on some cloud service. You should be able to still get a nice hot supper if the Internet goes down.
The app should show you a list of recipes that can be made with the ingredients on hand, you pick one or a couple, it tells you how long it's going to take to prepare. You tap go, and 45min later you have a perfect homecooked supper!
Literally everyone would get this. It would make the world so much healthier--enable everyone to easily diet, let people subscribe to healthy meal plans, etc.
It would pay for itself even at astronomical prices, via health benefits, no need to go out for dinner, and reduced food prices via making things from scratch.
Buying in bulk would be more common and much easier. So much comes from grain, sugar, salt and yeast. Much less food would be wasted--no need to batch cook anything and ignore the leftovers. The robot would be baking your bread, making donuts and croissants, tortillas, tortilla chips, mayonnaise, yogurt, sauerkraut, pickles. Things we never consider making on our own, that are generally considered the base components to make other things. I think it could easily offset $300-$1k in monthly expenses for just about everyone.
My assumptions are that it can be tailored to follow a recipe, of which many widely-published recipes are created/curated by chefs and experts, and once that is done the robot rarely, if ever, fails to produce the very best version of the dish.
If the only hindrance is just sitting down and writing the code, someone ought to get busy, because this is about as impactful of a development for the planet as any could be.
We can't even build a single-chore robot at the moment that replaces a human.
The software side of "chore-robots" is totally beyond anything that currently exists.
Think about the basic task of washing/folding/putting away clothes.
The AI would need
* Full 3D spatial navigation
* Be able to identify what is clothing
* Be able to reliably pick up clothing
* Be able to open doors to get around the house
* Somehow know how to operate every model of washer/dryer, including different possible placements of the machines
* Somehow know how to fold any clothing item
* Somehow know where to put the folded clothes (be able to open any kind of dresser?)
If you wanted to solve this you'd probably start with things like "well maybe there's a designated clothes bin that comes with the product that the robot understands" or "well maybe we make the user mark out exactly the space where they want the clothes placed"
But then you're leaving general AI territory at that point and heading back in regular robotics territory where the solution is to put the problem "on rails" as much as possible.
The "markets" where robotics could have orders of magnitude effects are what I call field to factory. The most obvious item is construction. A pre-fabricated house is cheaper and waaaay faster to build because the factory has robots and assembly lines etc to help. But even on-site construction can be assisted - that is turning the field into a temporary factory.
My sneaking suspicion is that just as we call any AI that works ML because while it does a specific job, we expect it to think like us instead of bringing what it is.
And for robotics we will have many amazing advances and social changes, but we will expect it to move like us so won't call it robotics.
This also leads to a second problem I think adoption will hit - robots will find an uncanny valley and it is easier and simpler to just build something that is in no way organic-like just to avoid human "yuck" factor. I can imagine some giant spider like contraption that washes up really well, and simply never selling because it looks like a giant freaking spider.
however thats not very forward looking of me.
We don't really need SCADA manipulators for clothes folding, we need a machine that we can just dump our washing in. Inside they might have a 6 axis arm, but I doubt it. There are machines that kinda do that for industrial linen companies. However they are huge and expensive, and you still need to feed in the clothes one by one.
Now what a decent robot arm could be used for is augmenting VR. Being able to tug, push, block and parry a person who's playing a sword game would be brilliant fun. Not without safety headaches, but fun.
I do agree that a decent arm thats cheap and powerful would unlock innovation. But they sorta do exist now. However we really need a usable IK library that doesn't require a maths degree to setup. We also need a library that would allow easy planning of moves as well. Its possible today, but hard.
Think trying to fetch and parse a JSON file over HTTP on a first generation arduino. Thats where we are with the libraries to make Robot arms usable.
Please no! Granting corporations control over your home physical environment is a dystopian scenario. Home hardware should be simple and reliable. The market has killed Juicero, and I hope that any company, that tries to pull this off, follows their way.
Maybe the killer app is selling "Robot Insurance" to people who are terrified by headlines like that.
I wish startup idea pitches were as clear as this essay.