Warehouses is definitely not where I expected robots with retractable blades to first appear.
The demo video: https://www.youtube.com/watch?v=oWXco05eK28
replies(6):
The demo video: https://www.youtube.com/watch?v=oWXco05eK28
https://old.reddit.com/r/EDC/comments/dmnuts/53mamazon_fulfi...
and Inbound (or the previous person picking) was usually a bit less careful.
There's a reason human beings are worked to the point of exhaustion in these warehouses - the goal is to move as much product as fast as possible. Quality and productivity are at cross purposes, and between the two only the latter makes money.
If five cheap robots outperform a single skilled worker, robots win. But depending on jurisdiction, those five robots might still lose to a dozen or so slaves kept near starvation. For the skilled worker it's bad news one way or the other.
Wait until LLMs get better and destroy the ability for junior developers to get their foot in the door.
A restaurant can improve performance during the "lunch rush" by letting neat slip, but that carelessness is already costing them performance at the end of the lunch rush - this works because just as this catches up they get several hours in the afternoon to clean things up. Then supper crowd where they do it again - then they have the rest of the night to clean up from that. (the restaurants I worked in didn't have a breakfast rush, YMMV)
A factory by contrast needs to keep things neat and consistent all the time because there is never a rush/downtime. They want things rolling off the line at a consistent pace all day. Any compromise for speed now is a cost latter in the day.
I have never been in an Amazon warehouse so I don't have great insight into what things are like. I would expect they want to be more consistent all day - but I don't know. Maybe all the trucks arrive at once and then they get time when they are gone to clean up. I wouldn't expect that, but maybe.
Yes some are better than others. However there is still a vast gulf in skill between those people than engineers (much less doctors), while the gap between them and someone off the street is much less. (the article doesn't say how long it takes someone to get to that high skilled state or even if it is possible to train to that level - if someone can show me data on this I might change my mind on skill)
From the article:
> “When you’re a person doing this task, you’ve got a buffer of 20 or 30 items, and you’re looking for an opportunity to fit those items into different bins, and having to remember which item might go into which space. But the robot knows all of the properties of all of our items at once, and we can also look at all of the bins at the same time along with the bins in the next couple of pods that are coming up. So we can do this optimization over the whole set of information in 100 milliseconds.”
Fillpy the robot will not:
- need vacations
- go on maternity leave
- call in sick
- steal from work
- be rude to customers
- go to work hungover from drinking
- come in high/stoned at work
- sue you for X,Y,Z
- sexually harass colleagues
- go on strike
- start a union
Look at the above restaurant example, the system has a built in buffer to handle spikes so it can be cheaper or make other tradeoffs everywhere else compared to an equivalently performant system that can do 100% duty cycle.
A robot or human that can deal with messy inventory is facilitating positive tradeoffs elsewhere in the system.
(The suits think that's a good thing)
The saying I do believe has a difference between robots and humans. The idea largely being that human inaccuracy increases exponentially relative to speed. Ergo, slowing down can lead to dramatically bette accuracy and throughput. Though, robots don't necessarily lose accuracy because they are moving more quickly. Though, I'd agree it is likely that both humans and robots need "smooth" in order to be fast. The key difference is robots do not always lose smooth when moving at high speed.
Several months of me as a doctor and I'd still be incompetent.
A robot that costs 5x as much but yields 30% more usable space is a better value proposition.
Also you wouldn't be able to accommodate anything besides items that fit in that specific pick bin (what a cubbie is called in the industry), meaning you would always need a near perfect match of # of bins of a specific size to the # of products matching them. It would be a continual battle to change racking as products move in and out of the warehouse.
A more complicated robot is a better robot, in this case.
Software has been an outlier in terms of its high salaries requiring only minimal training. That implies automating it will disproportionately be both easier and more valuable than many other skilled tasks.
While they're impressively good in each of the aforementioned three fields, they're still not the world-changing technology they were supposed to be, right? (At least, not by themselves. When a very powerful human ties themselves to a slop machine, believing the output to be real, this can change the world)
The main achievement, besides search which is a very useful application, has been in how effectively we can get real people to believe total bullshit.