Chopstick robot handle boxes with ease
spectrum.ieee.org
spectrum.ieee.org
Here's a commercially successful robot for the same job.[1][2] Universal Robots has some nice machinery. They've succeeded where Rod Brooks' Rethink Robotics failed. No cute display with facial expressions, but better mechanical engineering and short payback times. Also, UR's machines are safe around humans. The chopstick robot looks too dangerous to run outside a cage.
While AI for controlling robots in unstructured situations still isn't that great, the mechanical and electrical parts work much better. Motors and motor controllers are great and not too expensive. Sensors are widely available and pretty good. Compensation for flexing under load has made much progress, and robots no longer need to be as rigid as a milling machine. A decade or two ago, if you tried to do anything with robots, you spent most of your effort on those issues. (Been there, done that.) We're now past that.
Vision is far enough along that semi-structured jobs work pretty well. Pick up known object from bin, place in machine, get out of the way while machine does its thing, remove object from machine, put in output bin - that covers a lot of factory work. Especially with vision systems smart enough to notice that something is gone wrong, stop, and call for help.
I would assume that if it's anything like real chopsticks then the answer would be relatively little.
If you use chopsticks correctly, you can can use surprisingly little force to pick up realistically heavy objects. I believe the physics explanation of this ability is based around the concept of levers.
hmm, just realized, I've been on a kick to learn to do things left-handed so if I lose by abilities with my right hand I'll be ready. Or just becuase it's annoying to asymmetrical. So, I just realized if I learn to use chopsticks with my left hand, I will have 4 chopsticks at my disposal. I will be a force at the dinner table with the ha cheung.
As you say, slick/slippery being difficult comes down to chopstick skills.
The ultimate skills test is whether you can pick up something slick/slippery with Korean chopsticks.
Why Korean? Korean chopsticks are both thin and generally manufactured in polished stainless steel. Which makes them a greater dexterity challenge than the Chinese or Japanese formats.
I have a speaks-chinese assistant, and when we have asian lunches I always use use chopsticks and she uses spoons, and also thinks I'm crazy. By coincidence I just told her this story last week (because I didn't want her to think I'm an effete weebo or something) and afterward she mockingly bowed to me and said "thank you round eye for teaching me the old ways"
In the second video of the article the first box picked is crushed by the chopsticks.
Addressed a bit farther into the article, which I was thankful for because I had the same initial reaction:
“ You may have noticed that the boxes in the videos are pretty beat up. That’s because the robot has been practicing with those boxes for months, but Dextrous is mindful of the fact that care is necessary, says Drumwright. ‘One of the things that we were worried about from the very beginning was, how do we do this in a gentle way? But our newest version of the robot has the sensitivity to be very gentle with the boxes.’”
https://youtu.be/XPfWvqlEJ48?t=0m48s
https://www.universal-robots.com/plus/products/onrobot/2fgp2... The same paddle head can connect to multiple different arms and has suckers for sheets too.
I'm still looking for a good existing example of horizontal paddles with rotating wrist that competes with the chopstick idea. Most examples are restricted to known geometry, but I still think the chopsticks would be beaten by a pair of paddles every time (presuming similar control software).
Here's another palletizing arm and grippers combo:
I mean - imagine tying those chopsticks to your forearms and trying to manipulate boxes - it would be hard enough for a human. Much easier if tied two rubberised rackets to forearms?!
That was hard to parse. I was wondering how a "robot handle" would box with ease.
This doesn't really work when there is no conveyor to drop boxes onto, though.
But eh. Selling both is probably an obvious choice.
What really bothers me is the million dollar grant they got. Complete and utter waste of taxpayer money. Unbelievable. Most of us have to make real products to earn that kind of money.
[0] https://www.heil.com/bodies/liberty-automated-sideload-garba...
You just have to take whatever crazy spec you were given for the mechanical parts and then multiply that by 100 to engineer anything that could survive a single day.