Robotic Dishwashers And Dishwashing As A Service

There’s a story that goes back to the 1980s or so about an engineering professor who laid down a challenge to the students of his automation class: design a robot to perform the most mundane of household tasks — washing the dishes. The students divided up into groups, batted ideas around, and presented their designs. Every group came up with something impressive, all variations on a theme with cameras and sensors and articulated arms to move the plates around. The professor watched the presentations respectfully, and when they were done he got up and said, “Nice work. But didn’t any of you idiots realize you can buy a robot that does dishes for $300 from any Sears in the country?”

The story may be apocryphal, but it’s certainly plausible, and it’s definitely instructive. The cultural impression of robotics as a field has a lot of ballast on it, thanks to decades of training that leads us to believe that robots will always be at least partially anthropomorphic. At first it was science fiction giving us Robbie the Robot and C3PO; now that we’re living in the future, Boston Dynamics and the like are doing their best to give us an updated view of what robots must be.

But all this training to expect bots built in the image of humans or animals only covers a narrow range of use cases, and leaves behind the hundreds or thousands of other applications that could prove just as interesting. One use case that appears to be coming to market hearkens back to that professor’s dishwashing throwdown, and if manufacturers have their way, robotic dishwashers might well be a thing in the near future.


To be fair, the punchline to that professor’s challenge was probably somewhat undeserved. A dishwasher really doesn’t automate everything about the job of dishwashing. A grumbling teen still needs to load dishes into it, someone needs to add detergent, and the same grumbling teen has to be reminded over and over to empty the machine at the end of the cycle. So there’s still a lot of manual interaction, but it saves a great deal of work (and water) in the residential setting.

But what works in the home doesn’t work well in the food service industry. Restaurants and commercial kitchens face an entirely different scale of dishwashing problem than the residential user. Where no more than half a dozen place settings might be generated by any one meal at home, a restaurant might go through hundreds of plates, glasses, and bowls in an evening. And turnaround time is important — a residential dishwasher can sit with dirty dishes for a while, run overnight, and then be ready for the next day. A restaurant will use the same place setting for multiple seatings in a single night.

So there’s a lot of pressure on the one person in the commercial kitchen who is probably the lowest in the pecking order: the (human) dishwasher. Commercial kitchens generally have a (mechanical) dishwasher that blasts plates clean quickly with very hot, high-pressure water, but someone has to load and unload the machine, and that’s generally someone who’s about as happy to be there as the aforementioned grumbling teen. Aside from dealing with heavy piles of gross dishes and endless hours of standing while constantly being wet, dishwashers get little respect from the other members of the crew. That results in high turnover for those positions, which is a problem for management.

Robots for Very Constrained Washing Tasks

Enter the robotic dishwasher. A startup named Dishcraft Robotics is developing a unit for commercial kitchens that aims to reduce reliance on human dishwashers. The video below shows it in action, picking up plates from a stack with an articulated arm and placing them in a free spot on a rotating triangular fixture. The fixture rotates the dirty plate to face down so that a specially designed scrub brush can press up against it, scrubbing food away while jets of water rinse it. The scrubbed plate rotates to an inspection station where machine vision checks the cleaning job, and if the plate passes muster, it slides down a ramp so it can be picked up by a gantry arm and moved to a drying rack.

Now, there are plenty of bones to pick with this setup. First of all, it’s not exactly a human-free enterprise. Someone still has to move the stacks of dishes over to the machine, and wait staff no doubt does some kind of pre-cleaning of the plates. Aside from human factors, though, the demo is highly constrained. Dishcraft states that the dishes are picked up by the arms and held in the washing fixture by magnets, meaning the plates have steel inserts in them. This of course means that restaurants have to custom source their dishware, and fine china is probably not an option. That might be fine for more casual restaurants, but it certainly won’t fly with high-end restaurateurs.

There’s also a bit of ick factor from the fact that the bottom of the plate is not scrubbed of the debris picked up from the plate below it in the stack. Dishcraft dismisses that concern by saying the plate bottoms don’t get that dirty, and a spritz of water is enough to clean it off. I wouldn’t be too sure that a health inspector would agree with that assessment, so the machine may be in for some updates before this hits the market commercially.

Dishwashing as a Service?

Obviously, a system like Dishcraft’s is going to make financial sense only to the largest of commercial kitchens. That sharply limits the potential market for these robots, so Dishcraft is considering offering dishwashing as a service. Much like restaurants get all their table linens and uniforms cleaned by services specializing in the task, Dishcraft envisions restaurants piling their soiled dishes in racks to be picked up and swapped out for clean dishes for the next day. The dirty dishes will presumably be whisked to a facility where multiple robots will scrub them clean and return them to transport containers for the trip back to the restaurant.

On the face of it, this seems like a sensible business model, and one that investors will probably appreciate as it expands the potential market to smaller restaurants. But I see a glaring problem with this model, and it echoes the apocryphal professor at the beginning of this tale: Why would you build a fleet of robots to do what a bunch of dishwashing machines that you can buy off the shelf right now can do?

As a one-off in a busy kitchen, one can make an argument for the Dishcraft robot based on labor savings and fast turnaround. But shipping the dishes off to be washed takes away the advantage of turnaround time from the end user’s perspective. It seems to me that Dishcraft is a solution in search of a problem. Yeah, it’s a pretty neat idea, and kudos to the company for at least finding a way to automate a tedious and manually intensive job. But I don’t think dishwashing as a service is destined to be the killer app they seem to think it will be.

53 thoughts on “Robotic Dishwashers And Dishwashing As A Service

  1. If the delivery trucks were automated and there are enough dishes that a days worth can be buffered on site washing as a service could make sense. A large machine more like a dry cleaning facility could run 24/7 if sized to the local market. This would eliminate unproductive real-estate in the restaurant and allow for more thorough treatment of the wastewater from the washer. The carts should be redesigned to keep the dishes spaced as they are stacked, or it should be able to stretch, to allow the whole unit to be flipped sideways and run through the washer, since the dishes aren’t the only thing that gets soiled. This would eliminate the robot arms and individual handling of each dish. Bonus points for a gamma/x-ray/beta irradiation like is done with many foods before return of the unit to the customer to make sure everything is both clean and sterile. The more I think about it the better the idea sounds.

    1. But you’re forgetting the sheer weight of dishes and cutlery, there could be hundreds of pounds of ceramic and steel, and if your keeping dishes for all day and not constantly washing them, an average sized restaurant could need half a tonne of dishes, how is transporting all this practical?

    2. Storing that much clean dishes also needs a huge amount of real estate. Savings not to be seen. Why should an irradiation make sense, when you are able to clean the dishware with hot water? In the case this would not be enough (it is, since much more than hundreds of years), I would prefer to use hot steam to sterilize the stuff – no need for radiation safety. Irradiation is normally used for delicate goods like spices not for ceramics and stainless steel.

  2. As a teenager I used to wash dishes at a Chinese restaurant for 5 bucks per hour and free food. A robot will never compete on that level. And they will never be fast or flexible enough either. When you’ve got a Chinese cook with a large chopping knife screaming in your ear, you can clean something in seconds. Robots can’t do that.

    1. I’m not sure I’d want to place bets on that. Robots can be very specialised, and can move very quickly. Sometimes they don’t do an individual task particularly quickly, but they achieve a higher overall throughput. Additionally the cost of the labour isn’t the only cost, you have to take into account power, water, space, etc. A robot can conceivably beat a human system based on those factors leading to it being better value for money than the human.

      1. Imagine that Chinese cook shouting at the robot he wants two clean plates of a particular type now. Not in 10 seconds but NOW. Unfortunately, the robot doesn’t speak Mandarin and the plates are slowly making their way through the process that can’t be sped up. The cook hits the robot.

        This is EXACTLY the sort of scenario that would unfold at that little restaurant every night. That cook would slap me if i wasn’t quick enough.

        1. Sure, you and the cook would work together to make up for your limitations as a dish washing system. The actual good system wouldn’t have that limitation though, the cook would just take two clean and dry plates of the correct type from the storage area for that type of plate. The cook can still hit the robot if it makes him feel better though.

          1. Bollocks. All those “fast and efficient” robots are hopelessly lost outside their aseptic, shiny, perfectly controlled halls and assembly lines. The real world is eleven kinds of fucking messy, and you need prizes to get anyone to even attempt building a robot that can *gasp, shock, horror* open or close a valve on a pipe or open a door without falling over a dozen times – incidentally also why any video of anyone getting any small part of it working (hi there Boston Dynamics) automatically goes viral. Meanwhile, in the real world, we can’t even make a robot meant solely to crawl around and suck that, you know, doesn’t suck. Dream on and we can have this conversation again in, oh, fifty years or so. Right after we got fusion working, I reckon.

    2. There’s no chance that the iPhone is going to get any significant market share.
      Steve Ballmer, USA Today, April 30, 2007.

      After the rocket quits our air and really starts on its longer journey, its flight would be neither accelerated nor maintained by the explosion of the charges it then might have left.[1]
      The New York Times, January 13, 1920. The Times offered a retraction on July 17, 1969, as Apollo 11 was on its way to the moon.


    3. What is the quality level of the work of an underpaid teenager motivated by a scary chef? What is the difference of it on two different days, one month apart?

      If you pay for washing as a service, and the machine doesn’t work, you can sue the company. Can you sue an employee? Rarely, because they are protected by laws and rules.

      Consistency, reliability and accountability are more important for many, than cost. Robots CAN do that.

  3. If there is no control for conserving water on dishes that have easy-to-rinse food, it’ll be wasting water overall. I was a dishwater in college and we had to pre-rinse dishes, so we used less water on easier dishes before a rack of dishes got the 2-3 minute sanitizing process.

      1. Like last time after the barbecue: I put all the dishes on the floor (lawn) and used the smallest nozzle of the garden hose connected to a small pump (from the rain water barrel) so it gave me the highest pressure – up to 8bar – with lowest water consumption. Then I quickly “blasted” away most of the food residues and washed it with some detergent and some more water. But I avoided, having loads of ketchup, mustard or sauces in there.

  4. If you want to solve the home dish washing problem, you need two fridge-like cabinets. One holds the clean dishes and the other holds the dirty dishes. When one empties and the other one fills up, you close the door and start the washing cycle. That way you never have to empty the washer – you just switch between the cabinets.

      1. Normal dishwashers do dry the dishes. But it would be better to just have two identical big baskets for a single dishwasher: Put the freshly cleaned basket into the cupboard and the empty one into the dishwasher.
        Only requirement: Use all dishes before you need some cleaned

    1. I think some people do a variant of this with existing machines. Just keep the dishes you use most often in it all the time. After a meal put the dirty dishes back in and start the wash cycle. It would be interesting to see a wall cabinet mount for a dish washer. I imagine the plumbing and space lost to trying to hide the sump in a cabinet are the main reasons they are not mounted higher.

    2. Yes! When we built our house, I told my wife we’d save a butt-ton of money by just installing two high-end dishwashers and no cabinets. She didn’t go for it.

  5. It missed a spot on the back of the second dish. Since it just rotates there might not be a way for that particular tool to actually get that spot, regardless of how long it spins.

  6. The professor’s dishwasher story may be apocryphal, but there are plenty of real-world examples of this kind of thinking. There is a tendency to “teach to the test”. Older solutions to problems are ignored in favor of “how we do it today.”

    Endless examples show up every day. There are engineers that don’t know how to solder, or that resistors get hot, or that you don’t need a micro to do what a 555 can do. But they can write a C program in a flash!

    1. You are not wrong, though as much as I love the often elegant 555 circuits a microprocessor is usually a ‘better’ option now they are soo stupidly cheap. If the 555 circuit needed tuning for a particular task for example it could take board revisions or calibration pots in the circuit (add in calibration for each build) which is more time and cost than just throwing a cheap microprocessor on the board and altering the software it runs.

      And of course a Micro tends to be expandable to do other tasks when the use case changes – your 555 circuit might be brilliant reliable and cheap but it only does exactly what it was built to do.

      Does not mean it should not be part of the education though, along with the criteria to use. Much like for me far to many projects are all 3d prints when sheet stock could make that part faster and stronger if you learn how to work sheet Al/Acrylic etc. 3D printing is fantastic can do things you can’t do any other way but its not the only way to build.

      1. all too often I see 3d prints used as a “final” product rather than it’s intended purpose of prototyping. Is this acceptable? Eh…. in SOME very limited use cases maybe. But it’s far too often that 3D printed parts end up in the production BOM where injection molded components would be stronger and faster to produce, at least at scale.

  7. As someone who worked as dishwasher for exactly 2 months, I’m glad someone acknowledges how shitty the job is. I would have quit in 2 days if it wasn’t that the provided meals were excellent and the staff was friendly.

  8. What does this big machine that can malfunction, jam, requires additional cleaning and servicing actually improve in the dish washing process? Considering the plates need to be pre-rinsed by a human, this machine saves nothing. The only difference between this machine and a regular industrial dishwasher is you stack the rinsed plates into this vertical silo stacker instead of directly into a washing rack and on the output you get the same thing: clean wet dishes in a washing rack. However, as pointed out, this machine doesn’t even clean as well as a traditional washer as it leaves the backside dirty.

    I wonder how the suction end works when it encounters a large piece of lettuce or a bowl with leftover soup?

    This is a product engineered by people that have never worked in a kitchen.

  9. It seems like the real solution would be to make it not such a crap job?
    I’ve never worked in the industry, but Perhaps pay more for it? Some other perk to make it feel less of a raw deal? Or share the work between the team, so it’s not just the same guy, everyone does half an hour?

    1. The upside of the job is that they’ll take just about anyone, whether you’re a teenager who’s never had a job, have a terrible work history, just got out of prison, or have been laid off and want to have a bit of money coming in while you hunt for a real job. So there’s usually a decent sized pool of applicants, and since restaurants have pretty high turnover at all levels, there’s also opportunities to promote the good ones who want to stay.

      1. Yes, that’s true. Human population is not all engineers (or think any other high-education job). The others also need jobs. Only sitting around and drinking is no life for most people.

    2. More money won’t really help if the work is crap. To make it not suck the work conditions need to be decent – proper temp (not a sweatshop), give employees padded mats or nice nonslip sneakers, not RSI-inducing work, reasonable work/time pressure, reasonable hours, management that listens and makes changes, etc. IMO crappy jobs are usually crappy because of management decisions or neglect. Have all that sorted and then you just need to find somebody who likes the zen of working, which isn’t so hard.

  10. Apparently most of you have not actually been in a commercial food operation for say 600 to 1000 people 3 times daily. So perhaps 3000 to 5000 operations a meal just for plates and cups. Double that with silverware. You don’t want spinning or rubbing parts that either break, jam up, or need constant maintenance. Just water jets at high pressure are enough of a maintenance pain. Go improve a commercial Hobart.

  11. It would be much easier to automate picking and loading of dishes and flatware into the racks for current commercial dishwashers. Wheel in a cart stacked with dishes from the dining area, the robot picks the dishes and gives them a shake over a trash bin (perhaps with a bit of air jet) to knock off any large food remains and other waste like paper napkins and straw wrappers. Then it would give the dish a quick up and down in a water spray, then load into the dish’s designated type of rack slot. When the rack is full, the sides of the washing cabinet raise, the rack full of dirty dishes and flatware is pushed in, pushing out the rack full of clean dishes. The other robot picks and stacks the clean dishes, ready for plating food by the cooking staff. The empty rack gets cycled back to the dirty side.

    It’s exactly the same process as is commonly used in many commercial and school kitchens, but replacing one or two humans with a couple of robots. There are end effectors designed to handle odd sized, weighted, shaped, hard, and soft items. There’s also machine vision systems capable of picking out and sorting objects that are overlapping and/or intermixed with other items. In case of flatware missed and dropped into the trash bin, that could be automatically recovered by having the bin emptied via a chute with a magnet trap into the food waste dumpster outside. That’s a thing already in use in pellet mills that make wood fuel pellets and animal feed. Have to pick any ferrous metals out of the material stream so it doesn’t get into the mill and damage the dies.

    Years ago I saw a video on new places robots were being used. One was at a pastry factory where items had to be transferred from one roller conveyor to another, set at some odd angle to the first, and the items had to be flipped over while being transferred. The human worker used a very long and narrow spatula to slide under a row of items then in a precise motion lift, pivot and flip simultaneously. What the factory *could have done* was have a bespoke connecting unit built to slot exactly between the two conveyors to turn the corner and flip the items. What they did instead was buy a general purpose multi axis robotic arm and affix the spatula to it’s 3 axis end joint. The robot was programmed to exactly duplicate the hand and arm motions of the humans who did that job.

    The benefits of that robot, despite costing as much or more than having a custom turn and flip transfer unit built, was that if the company reconfigured their equipment, the robot could be used between any two conveyors and could be reprogrammed for the new arrangement, and to use different tools. The robot could lift, turn, and flip pastries around the clock, in the dark, without ever needing a break or having to tag team handoff with another to keep the line moving.

    The ABB FlexPicker which came along some years later couldn’t handle a job like that, not without some extra device to turn the pastries over. One of those would have to transfer the pastries one at a time, at high speed, and would have to stagger placement in order to keep the lines straight.

    That makes two main design cases for automation. Adapt a robot to an existing system to replace humans or design a whole system to work with specific robot designs like a FlexPicker which has as its main strength very rapid relocation, with or without horizontal rotation, of a large variety of items.

    A fully automated cook to order fast food restaurant would likely use both methods. Multi-axis robot arms to work with adapted existing food equipment, and newly designed equipment made to work with standard high speed product handling robots with specialized end effectors.

  12. This is why conveyor dishwashers exist. Many don’t even need you to load a rack or stack things. Just place them face down on the infeed. Collect after dried.

    Thisnisna case of overcomplicating a process for the sake of ‘simplifying and saving time’s yet not actually solving any of the time consuming steps. Who’s going to perfectly stack dishes?
    And can it support a wide variety of plates, bowls, and cooking utensils?

  13. A) Have the designers ever worked in a kitchen? Do they honestly expect servers to neatly stack dishes in the appropriate piles to not mess up the robbit?

    B) Have the designers been outside of their computer lab, ever? That system has so many moving parts to break down and no, the machine will not be properly maintained, it’s located in a commercial kitchen. Current machines that have a pump, a door and a heating element break down permanently once a year, how is this juicero going to do in comparison?

  14. So none of have ever worked as a dishwasher, because the dishwasher does many things besides wash dishes that robots will never be able to do. Number one and most important is physical presence to assist the servers when an unforeseen event happens, like a big spill, or a backed-up toilet or an unruly customer who needs to be ejected from the premises. Number two is the psychological presence to come to the aid of a waitress who has been sexually harassed or a cook who has been unjustly criticized for making food as instructed.

  15. Dish washing as a service has it’s place. When I went to China on a business trip, I ate with my coworkers at a restaurant where the dishes were waiting at the tables shrink-wrapped in small sets, with a plate, bowl and tea cup wrapped together. I was curious about this, and my chinese coworkers explained that the dishes are washed at a separate facility, shrinkwrapped and sent to the restaurant. I doubt the dish washing is fully automated though.

  16. The only thing which is needed here is a sound-damped pressure washing cabinet with a washing pump which can supply also hot steam, a clear view screen like on ships of locomotives ( and a set of insulated long sleeve rubber gloves (like on sandblasting cabinets, but insulated).

    And there has to be just one thing automatic on this cabinet: “Clean&Disinfect&Dry yourself when I’m done washing something.”

    – Pressure washing/Rinsing off dirt/Sandblasting is one of the most satisfying activities.
    – It would be way more versatile: this cabinet could also clean off the most burnt, crustiest of oven pans, dirty rubber boots, ugly sneakers, kitty litter boxes, each and every scubby bucket, bowls with dried-on dough, and if you include a soft wash/rinse only program, even puppies which rolled themselves in manure or toddlers from the sandbox.
    – How many things did you throw away in the last 12 months only because you considered then uncleanable?

  17. I don’t think the people who designed this robot ever worked in a kitchen or any other real world scenario.
    I can see the vacuum system getting fouled by food and dishes dropping and breaking inside it.
    The numerous surfaces would tend to catch debris and breed bacteria and the transparent plastic parts simply will not hold up to industrial use.
    There’s a reason why food service equipment is made from stainless steel.

  18. I am like Three years late to this… but I’m glad to see that my videos are being used as a demonstration. Now for the main topic at hand… Robots could be awesome to an extent when it comes to dish washing. Every restaurant is different and every dish pit is made different in each restaurant. Robots wouldn’t replace the dishwater because he or she would have to maintain the robot to do the tasks to fit the restaurant’s Layout.

  19. it’s easy to come up with automated process (like a robot) to wash dishes in a commercial/restaurant/hotel businesses. In such places, you have plenty of room and power without nose restriction. The challenge here is that, with 130 million household in the US alone, we need a practical dishwasher to be used at home. Well folks, I did that with AI. With my invention, you will never have to load and unload the dishwasher. If a dish is not clean, it will go through additional cycles until AI says “It’s enough!” which it will then discharge the dish. Here is a video of the prototype:

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.