(50X) Autonomously folding a pile of 5 previously-unseen towels
RLLberkeley0 HN recommenders · 2014–2024
923K views · demo · dated · Uploaded 2010-03-17
A Berkeley PR2 robot autonomously folds a pile of five previously unseen towels, shown at 50x speed. The video accompanies a 2010 ICRA paper from Pieter Abbeel's group.
HN cites it as the benchmark for how hard and slow robot laundry folding was, in threads about household robots and AI progress.
15:01What Happened to Horses Is Happening to Us
CGP Grey
11:46Zipline - How Rwanda Built A Drone Delivery Service
Real Engineering
21:32Amazing Invention- This Drone Will Change Everything
Mark Rober
5:087 Minutes of Terror: The Challenges of Getting to Mars
NASA Jet Propulsion Laboratory
7:48Slaughterbots
Stop Autonomous Weapons
18:40Drone Delivery Was Supposed to be the Future. What Went Wrong?
Wendover Productions
4:02Giant Robot Project, Part 1
Jaimie Stuff
7:59Sci-Fi Short Film “Slaughterbots” | DUST
DUST
10 comments from Hacker News
[VIDEO] Work is progressing. That's a $300K robot. (edit: no affiliation. Video shows PR2 robot at Berkeley folding towels competently but very slowly in 2010)
This article has issues. XKCD explains it best here: [LINK] Take for example... "Other low-wage and labor intensive jobs in retail, such as stocking the shelves of supermarkets with food, will soon be replaced by machines that can do those jobs better and faster than humans could." Here's a state-of-the-art robot folding some towels. [VIDEO] The amount of image processing and calculation needed to fold towels or stock shelves is in fact well beyond the reach of current AI / technology. Yes yes yes, there's that shirt folding machine prototype going around. But that only gets the job done with a good deal of human assistance. (Placing the shirt on the machine correctly, assuming shirts are a certain shape, etc. etc.). A "Shelf-stocking" robot sounds like a horribly complicated AI task, the likes this world has never seen before. A research team might be able to accomplish it, but such a thing is clearly beyond the reach of typical tech within the next 10 years.
Here's the same guy (PR2) in action 7 years ago folding towels at 50x speed: [VIDEO] And I doubt it was trained. It seems to be following a predefined procedure instead of learning by itself. It seems to have come a pretty long way if you ask me!
It is entirely possible [VIDEO] , but there is no market for a domestic robot. Not even for an AiBo. You see: poor people are accustomed to doing these chores by themselves, while the rich can always use their paper money to hire naturally occurring biological robots.
You forgot the wonky '60s-look robot solution: [VIDEO] (Actually, pretty cool). Your industrial solution reminded me of the time I worked at a factory feeding books into conveyor belts. I guess I'm weird so I kind of enjoyed the sensation of my fingers and my hands learning to grab exactly the right number of catalogues from a cardboard box, or exactly the right page to open a magazine on to stick a cd in etc, after I'd done it a few dozen times already in a day. And at least I'm grateful I only had to do one set of repetitive moves for each task (rather than the whole symphony of manipulations the machines performed). But, man, is that sort of work inhumane and soul-destroying. I'd have been soul-destroyed myself if I had a soul. (No, seriously, that sort of work sucks so much)
That's probably because you already have a robot that does 90% of your dish washing. A dishwasher. But none of the loading and unloading yet, unfortunately. Folding clothes is super hard: [VIDEO]
If the the lowest hanging fruit is a loosely structured hi-DOF dextrous manipulation task we gonna lose our shirts before we fold 'em. Laundry is the domestic equivalent of de-mining: the perfect robot task if it weren't for the technical and financial challenges. You have a laundry robot at home already and it does everything except folding for a reason. Here's a video of the state of the art not too long ago, using a $500K robot. Video is 50X speed. [VIDEO]
Ever since I came across this video [VIDEO] of a robot taking 1,5 hours to fold five towels, with mixed results, I think about how much of our brain might be devoted to handling fabric every time I have a towel in my hands. Probably the same with handling fluids, like predicting how milk is going to behave in an opaque carton.
Doing these cute demos is a double-edged sword. A fun and charming demo attracts attention by non-experts who can't see the wood for the trees. In my field we make robots do a thing, say folding laundry, as an example of dextrous manipulation[1], and the popular conversation is about whether your laundry robot makes commercial sense and not about your novel grasping strategy and nifty motion planner that makes this sorta-possible for the first time. But everyone sees your video. Pros and cons. [1] laundry folding at Berkeley (not me) [VIDEO]
We are certainly getting close! In 2010, watching PR2 fold some unseen towels is similar to watching paint dry [1], but we can now enjoy robots attain lazy student-level laundry folding in real-time, as demonstrated by π₀[2]. [1] [VIDEO] [2] [LINK]