Dishcraft Robotics Takes Over Dishwashing From Humans

Washing dishes is a problem that robots can solve, at least in commercial kitchens

5 min read
The Dishcraft Robotics dish washing robot system uses a robotic arm and AI
The Dishcraft system cleans each dish individually using a rotating scrubber head. It then inspects each dish in a fraction of a second for remaining food residue using computer vision and machine learning to achieve a consistent level of quality.
Image: Dishcraft Robotics

The kinds of jobs that robots are ideal for are the kinds of jobs that humans just straight up do not want to do. This is where the whole “dull, dirty, dangerous” thing comes in, but even in those categories, some jobs are duller, dirtier, or more dangerous than others. These are the jobs that we should be focusing on robotizing—not just jobs that are possible to automate, but jobs that need to be automated because you simply can’t find enough humans to reliably do them.

One of these jobs is commercial dishwasher. It’s dull and dirty, and turnover is very high, with the average human quitting after just over a month and around 30 percent of dishwashing jobs going unfilled, according to one estimate. And if your dishwater doesn’t show up for work, everyone else in the kitchen has to pitch in to make sure that there are enough clean dishes, slowing everything down. 

Today, a startup called Dishcraft Robotics is announcing a new robotic dish cleaning system, designed to minimize the time and effort that humans spend scrubbing dishes. Brought to you by some of the folks behind Neato Robotics and Dash Robotics, the San Carlos, Calif.-based Dishcraft uses some clever engineering and practical constraints to make sure that dishes are done cleaner, faster, better, and cheaper.

First, some context: There are, of course, already robots that wash dishes, and you may have one in your house. But a home dishwasher doesn’t hold enough dishes and takes far too long to be useful in a restaurant kitchen. Solving this problem is not just a matter of cranking up the speed of a home dishwasher—part of the reason why a home dishwasher functions at all is that it replaces human scrubbing with water, heat, and time. Nobody has time for that in a commercial kitchen, though, which means that scrubbing is necessary. 

Getting a robot to scrub dishes is a really challenging problem if you try to do it like a human does, because it involves grasping an assortment of very slippery things and then manipulating them with a substantial amount of dexterity. Robots are not ready for this on a commercially-viable scale, so Dishcraft had to find ways of making an automated system that’s a compromise between what robots can realistically do and what a kitchen needs. Here’s what they came up with:

The Dishcraft system inspects each dish in a fraction of a second for remaining food residue using computer vision and machine learning to achieve a consistent level of quality. But the really clever bits here are how tightly Dishcraft has constrained the system. First, dishes are pre-sorted and pre-stacked into those carts, solving two big potential problems for a robot thanks to humans for whom the tasks add only a very minor amount of effort.

Once the carts are wheeled into the Dishcraft system, you’ll notice that there’s a robot arm that picks up each dish one at a time and moves it over to where it gets cleaned. The arm isn’t using suction—the dishes are all custom, made with a bit of steel built into the bottom so that they can be picked up magnetically. It turns out that many places don’t actually care what dishware they use, so if Dishcraft can provide them with new dishware that’s much easier to clean, then great.

Each dish is individually cleaned with that rotating scrubber head, which is designed specifically to work with the range of dishes that the system is expecting. It took Dishcraft a lot of experimenting to find the scrubber head that works best, and they sent us a picture of some of their iterations:

Dishcraft Robotics scrubber evolution The Dishcraft scrubber head evolved over several iterations until the company found the design that works best. Photo: Dishcraft Robotics

After the dish is scrubbed, it’s rotated through 120 degrees where it’s examined by the vision system for cleanliness. If it fails, it just rotates back around to get cleaned again, but if it passes, it’s placed into a kitchen-standard dish rack and slides out of the system. From there, the rack passes into a sanitizer (which most commercial kitchens already have) for the final cleaning step.

The overall idea is that you barely need any human involvement at all, and certainly no active or repetitive work on the part of a human. There’s putting carts into the Dishcraft system, and then taking trays out of the system and putting them into a sanitizer, and that’s pretty much it. You don’t really need a dedicated “dishwasher” anymore, and perhaps someone who’s doing something more interesting could spend a little bit of their time minding the Dishcraft system as well.

Dishcraft Robotics uses custom dishes, pre-sorted and stacked Dishes are pre-sorted and pre-stacked into carts before they got into the Dishcraft machine for washing. The dishes are also custom, made with a bit of steel built into the bottom so that they can be picked up magnetically by the robot arm. Photo: Dishcraft Robotics

As constrained as Dishcraft has tried to make things, there are still humans in the loop, which means that they’ve had to make sure that their system is adaptable. For example, it works best if most of the food has been scraped away already, but it’s been designed to be able to handle stray t-bone steaks that have been left on plates (using an updated end-effector that isn’t shown in the video) that can make multiple passes if necessary. It can also deal with old, dried up food, or things like nested bowls that had mashed potatoes in them causing them to stick together. After working on the problem for three years, Dishcraft is confident that the system is now ready to go to market.  

Many kitchens, it turns out, don’t need their own Dishcraft robot. What they do need is clean dishes, so Dishcraft offers “dishes as a service,” exchanging clean dishes for dirty ones and then washing them in a centralized location. This is the same model that most restaurants use for linens, like tablecloths and napkins. The company is also doing custom installations for places that need more throughput.

Dishcraft was founded in 2015 by Linda Pouliot and Paul Birkmeyer, both robotics veterans: Linda co-founded Neato Robotics, and Paul co-founded Dash Robotics. So far, Dishcraft has raised a healthy US $25 million, which is not all that surprising—they’ve managed to identify a real problem that can be solved right now by a robotic system in a reliable and cost effective way, and what they’ve come up with seems quite promising.

[ Dishcraft Robotics ]

Update- we had a few people ask how the bottoms of the dishes are cleaned, so we checked back with Paul at Dishcraft about it. Turns out that it’s tricky, since they only scrub one side of the dish, but in testing they’ve found that dish bottoms don’t get the same kind of stuck-on gunk, and some well-placed water jets can get the job done. “We have found great success in doing a bit of a choreographed dance between nozzles that target the back of the dishes and a movement of the dishes themselves back and forth away from the surface of the wheel,” Paul says.

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman
LightGreen

This article is part of our special report on AI, “The Great AI Reckoning.

"I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less