Their sales pitch seems to be roughly that the value is more in the machine learning software than the gripper itself. I assume having both means the software can choose how much of each (fingers/suction) to use for a particular item, after it identifies the item.
So perhaps it's supposed to work best when there's a variety of items versus a few predictable ones.
while certainly cool, i have always found such robots disturbing because instead of being used to sort recycling or some other useful purpose, they are meant to actually increase the disposal of plastic and other such waste by moving product at ever increasing rates.
This is pretty neat! I think we are so predisposed to assume that grasping and lifting things should look a certain way based on our own anatomy.
The big questions that come to mind are resilience or I guess you could call it MTBF, and also hygiene.
A human hand knows when a thing it picks up has gotten it dirty and can change a glove or go wash. This design in particular looks like it would hang on to particles from things it’s picked up (thinking like bits of broccoli, but also on the microscopic level) and be very difficult to clean with all that surface area.
I think for picking groceries you would almost have to dedicate one arm for each SKU placing items into bags that were ferried by, you could not have one arm with a bag moving between SKUs and filling that bag.
I fully expect to see lights out grocery stores in the near future, I’m a bit surprised no one has tried it yet. Maybe the last mile delivery still dominates the cost too much.
> I fully expect to see lights out grocery stores in the near future, I’m a bit surprised no one has tried it yet.
Do you mean fully automated? I think it's the pick/place of irregular objects that's the problem.
In the UK, grocery deliveries are pretty cheap and available widely.
I actually just looked up ocado to show you a video of the human packers and found that mid last year they were running a fully automated warehouse, and then that it burned down about a month ago.
> I think we are so predisposed to assume that grasping and lifting things should look a certain way based on our own anatomy.
Yeah, that is very interesting about it. Also interesting is that it is an endoskeleton with a soft outer hull that is more similar to flesh than the hard rubber cover that you sometimes seen on "traditional" robotic grippers.
One could describe this device as replacing eye-muscle and (position, pressure) sensor-muscle feedback with cleverly engineered mechanical compliance. So much for solved problems in motion planning...
I don’t think it would be difficult to clean or change a rubber lining. It looks like it would be smooth when removed. Definitely it would need to be used carefully around difficult to remove allergens.
Interesting point of view. Perhaps using a Teflon + antimicrobial coating for its "skin" and integrating a washing device in the robot for it to clean its gripper every now and then could work?
Maybe make it transparent and have some UV led's inside to kill microbes. Occasionally make it grip an air blower to remove dust. And at the end of the day put it in a dishwasher!
A piece of string can lift 100 times its own weight. The interesting thing is that it can lift a reasonable variety of different things delicately, in a reasonably simple way.
Impressive, but I'm not sure there aren't some class of objects that will give it a hard time. From The videos it seems that it needs to engulf the object and to do so the object must small enough and be well oriented. I'm not sure it will pass the remote or smartphone on the table test.
I'm going to crack up if the most effective general robot hand is a flap of rubber. One potential benefit is those handthings won't be able to fire a gun or wield a sharp weapon very well. I half joke but the robot maids are probably going to be quite hackable.
