When will we see the first home dishwasher cobots?
15.14, Wednesday 25 Aug 2021 Link to this post
Sometime while I wasn’t paying attention, robotics got really good. But mainly in industrial settings. So beyond Roomba, how do robots come into the home?
The trend I’m tracking is cobots - collaborative robots - originated by James Colgate and Michael Peshkin at Northwestern University in 1996. From their homepage, a cobot is
a robot for direct physical interaction with a human user, within a shared workspace.
The cobots patent makes the supporting role clearer:
cobots guide, redirect, or steer motions that originate with the person.
Although “cobot” is a general approach, the typical approach seems to be a robot arm with sufficient sensors to avoid workplace injury.
For example, here’s Universal Robots explaining the benefits of cobots on assembly lines:
- They’re easy to “program” (no-code style) and take over precision work but not the decision-making:
you simply move the robot arm to desired waypoints
- They’re flexible and easy to re-task:
Moving the cobot to a new process is fast and easy, giving you the agility to automate even small batch runs.
- They’re safe:
80% of the thousands of our robots worldwide operate with no safety guarding.
The difference in approach between regular robotic automation and a cobot is fundamental. It’s the familiar promise of automation, but without having to move to a 100% automated production line. So you could imagine, putting up a fence, you could have a cobot lean in to just punch in all the nails in all the right places while you hold the plank in place (which means your team size is reduced to only one human), but you don’t need to swap out to a massive fence-construction-machine.
ALSO fascinating is that these cobot snake arms generally come as general purpose platforms for which you develop “apps”. So…
- Here’s a platform: Sawyer from Rethink Robotics. It’s a 7-jointed robot arm with a gripper on the end, collision detection sensors, computer vision, and a network connection.
- Here’s a demo “app”: Cobot Cafe (YouTube) which is Sawyer functioning as a barista, remote controlling the coffee machine to pour the coffee, and moving the cup with its arm.
I’d like a cobot at home. They’re safe and portable, that’s the promise.
So maybe I could place my safe, portable cobot in the floor in the front room, and it would pick up all the toys and tidy them away, shelve any books, and find the TV remote control and put it back in the regular place.
What I like is that we inhabit the same space, and I think of the distinction like this:
- I have a robot to clean the dishes already. It’s called a dishwasher, and I have to make room for it, and adapt my behaviour to use it (loading and unloading). But despite having the dishwasher machine in my kitchen, I still need the sink, and I still need a sponge and the Fairy liquid, and so on, because I wash pots and pans like that.
- But a cobot dishwasher would stand alongside me at the sink. I would stack up the dishes and so on, and it would clean them and pass me the items to put away in the cupboards. I would get back the room previously devoted to the big machine. The cobot would use the exact same cleaning utensils as I do – and I could move it to go and do other jobs afterwards.
While we’re on dream apps for domestic cobot arms:
I open my (physical) post approx every 12 months. I pile it up over the year. It’s a pretty mechanical process to recognise each item, sort and stack it, and to pile up the recycling. If it could read the addresses, show me each unrecognised item (remembering my response for later), and run the batch process, that would cut a good few hours out of preparing my tax return.
How will the cobot be domesticated?
The mainframe was domesticated as the personal computer, and the PC was dismissed as a toy – and then came spreadsheets and then came desktop publishing, killer apps both.
So what’s the first home-use programmable cobot that seems ludicrous to begin with, but establishes a beachhead?
Part of me wonders whether it will be cooking: kneading dough, sitting by the stove to flip pancakes, using sensitive fingers and eyes to cook the perfect steak, and so on. It could even wash its own hands.
But the kitchen is a pretty inhospitable first environment.
Perhaps the first use is for hobbyists? Imagine painting miniature figures – hey arm, hold this; pass me the cobalt; rinse this brush and put it back. Or fixing a bike. Or assembling Ikea furniture.
None of these routes feels quite believable to me. But my takeaway is that the blocker is market entry. The technology is all there. If there are teams working on consumer smart glasses in readiness of them becoming commercially viable (which there clearly are) then I hope there are teams working on consumer cobot arms.
I subscribe to a small handful of blogs which track the latest in robotics (thank you RSS).
- Robotic Gizmos
- Soft Robot Critics (on Tumblr)
- Iskander Smit’s consistently excellent newsletter Target_is_New
Also a shout-out to Robin Sloan’s 2018 novel Sourdough which, behind its deceptively gentle facade, is an exploration of human/non-human cooperation and where agency lies, whether that’s yeast or - relevant here - robot arms. (Or even machine learning, given that’s how Sloan collaborated to compose the fictional music of the Mazg for the audiobook.)