Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

How Amazon Triggered a Robot Arms Race

bloomberg.com

Post reply on HN
Print view View original

81–90 of 103 posts

Re: How Amazon Triggered a Robot Arms Race

#81

Post by apendleton »

ethbro wrote: post #61
sbov wrote: post #11

Earlier quoted context omitted.

No, all you need are cheap robots to get past that local maxima. The point of UBI is to ensure people don't become destitute when we reach that point.

Robotic automation of any sort doesn't occur in a vacuum. If 50% of the jobs in a field are automated, in the near term you still have 100% of the former labor looking for that job, pushing salaries down towards their minimum, thereby reducing the cost effectiveness of further automation.

Though of course (without passing a value judgment on such policies) minimum wages set an artificial lower bound on the cost of human labor here, and this kind of low-skill work is inexpensive enough that hitting that bound is likely (vs., say, automating legal work), especially as minimum wages rise. For some industries getting to the point that near-100% automation is cheaper than $15/hour for humans is totally plausible to me.

Re: How Amazon Triggered a Robot Arms Race

#82

Post by bedhead »

sgwealti wrote: post #75
endersshadow wrote: post #73

Earlier quoted context omitted.

What's an example of them? I'm interested to learn.

There are 5 or 6 vendors that have offerings in that space. Some of them are: Opex Perfect Pick, Dematic Mulishuttle 2, Intelligrated OLS Shuttle, Witron's OFS, SSI Schafer's Navette. These solutions are definitely more capital intensive than Kiva robots but they are faster and have a much smaller footprint. If you have a Kiva-type goods to person robot system and your order profile changes like more lines per order…

Isn't the issue that a Kiva system is more scalable since it allows for rapid handling of far more SKUs? A key part of Amazon's retail strategy is that it's the first place you go to look for something because you know they have seemingly product on earth...they're the default e-commerce site for many. My understanding was that these shuttle-type systems simply don't allow for that. Maybe I'm wrong...if so what are the other trade-offs? Thanks

Re: How Amazon Triggered a Robot Arms Race

#83

Post by igk »

kristofferR wrote: post #24

Komplett.no, which aims to become the Amazon of Norway (the real Amazon doesn't care about Norway), has a really cool automated warehouse (AutoStore) which seems even more advanced than Amazons: http://autostoresystem.com/ https://www.youtube.com/watch?v=_O0UKARdLDc (360° video) http://www.tu.no/artikler/unik-video-bli-med-inn-i-kompletts... The packages still need to be packaged by hand though, but it seems like it…

I just interviewed at autostore:) They seem have really cool and clean technology and the one person I had contact with so far seemed really nice.

Re: How Amazon Triggered a Robot Arms Race

#84

Post by antisthenes »

kuschku wrote: post #25
Animats wrote: post #18

I'd wondered how Kiva's old customers were doing. Now we know. There really isn't a direct replacement for Kiva robots from another vendor. The great thing about Kiva robots is that they're quite simple and robust mechanically. They're a battery, motors, wheels, and a lift mechanism. They're completely interchangeable, and having one fail isn't a big deal. Maintenance is replacing batteries, motors, wheels, and maybe…

And this is horrible for the market. A company gaining a huge advantage by getting a total monopoly over another market, which they can enforce due to having patents. That’s not what patents were supposed to create, and what Antitrust regulations were supposed to prevent. If this is the future, it’s quite a dystopia: Monopolies everywhere. Sure, I get that the Silicon Valley business model is to destroy regulation, g…

Properly regulated monopolies are the most efficient form of a market.

In fact it is quite the opposite of what you're imagining.

Re: How Amazon Triggered a Robot Arms Race

#85

Post by antisthenes »

brc wrote: post #52
lhopki01 wrote: post #43

Earlier quoted context omitted.

Monopoly is the natural end state of capitalism. Once a company is big enough it's easy to destroy or buy out all competition. It's only regulation that prevents all industries becoming monopolies over time. You're confusing monopolies coopting government to maintain control with large government causing and maintaining monopolies.

Sorry, but no. Creative destruction is the end state of capitalism as has been proven time and time again. There are no examples of monopolies lasting more than a few years unless they have overt support from the state. Large governments cause monopolies to happen. You said the same thing. Business co-opting a government to creat a monopoly is only possible when the large government exists. You could argue that water…

> The ultimate cure for monopoly behaviour is a smaller government and more competition by avoiding regulations that favour larger operations over smaller ones.

That would be the libertarian fantasy, but no, unfortunately that is entirely incorrect.

The "regulations" that you assume make monopolies possible is really unrestricted capitalism in itself, which leads to accumulation of capital in the hands of one or few, often colluding entities.

Capitalism favors larger operations over smaller ones through efficiencies of scale. There is no human-created regulations that do this.

Re: How Amazon Triggered a Robot Arms Race

#86

Post by antisthenes »

franK_ wrote: post #16
noonespecial wrote: post #12

Earlier quoted context omitted.

That people "need" awful picking jobs they hate doing but can be done better by robots is basically the existential problem humanity needs to solve this century or we are very likely to get Filtered, Robin Hanson style.

It's only existential in sci-fi books. Textiles were one of the first industries to be automated with the punched cards\power loom 150 years ago. We still have millions of human weavers all around the world. The problem with automation is you cant ask the robot why X, Y or Z is going wrong and how to fix it. A master weaver on the other hand will do it over a 15 minute conversation.

> We still have millions of human weavers all around the world.

Yes, mostly in Asia, living in hellish conditions, often struggling to get enough food. I'm not sure what point you were trying to make with that comment.

Re: How Amazon Triggered a Robot Arms Race

#87

Post by sgwealti »

bedhead wrote: post #82
sgwealti wrote: post #75

Earlier quoted context omitted.

There are 5 or 6 vendors that have offerings in that space. Some of them are: Opex Perfect Pick, Dematic Mulishuttle 2, Intelligrated OLS Shuttle, Witron's OFS, SSI Schafer's Navette. These solutions are definitely more capital intensive than Kiva robots but they are faster and have a much smaller footprint. If you have a Kiva-type goods to person robot system and your order profile changes like more lines per order…

Isn't the issue that a Kiva system is more scalable since it allows for rapid handling of far more SKUs? A key part of Amazon's retail strategy is that it's the first place you go to look for something because you know they have seemingly product on earth...they're the default e-commerce site for many. My understanding was that these shuttle-type systems simply don't allow for that. Maybe I'm wrong...if so what are t…

With either system you can scale in the following ways: -Add more shelving units to store more SKUs or to store more of faster moving SKUs. -Add more robots to process more order lines simultaneously. -Add more pack stations/labor

Adding more shelving units is limited by floor space. The equivalent scaling for shuttles is adding more aisles. Adding more aisles scales farther than adding more shelving units because it uses the entire vertical height of the building. Even taking mezzanines into account aisles are going to scale farther.

Adding more robots or shuttles will scale until the travel paths are congested. Whichever technology moves faster is going to scale better here. I don't know enough about the speed of the Kiva robots vs shuttles to say who is better here but I suspect shuttles can move faster.

Adding more pack stations is something either type of system can do and I can't see any advantage either system has over the other there. Kiva is probably more flexible here because you don't need expensive conveyor reconfiguration/addition if you want to add more pack stations. Shuttles, with the proper supporting conveyor can probably get higher throughput.

I think where Kiva really shines is in its flexibility. It can store a lot of different types of items and it is flexible in its ability to move stuff around. It is definitely not the most labor efficient goods-to-person automation technology and it is also not a great use of space. It runs into problems if you need to scale up in SKUs per order or overall order lines per day because you don't have enough travel space available to get all that stuff to the pack stations. Based on my ordering habits and what I have seen of friends Amazon is dealing with a very low number of SKUs per order so it works for them.

Re: How Amazon Triggered a Robot Arms Race

#88

Post by icebraining »

ximeng wrote: post #64
brc wrote: post #52

Earlier quoted context omitted.

Sorry, but no. Creative destruction is the end state of capitalism as has been proven time and time again. There are no examples of monopolies lasting more than a few years unless they have overt support from the state. Large governments cause monopolies to happen. You said the same thing. Business co-opting a government to creat a monopoly is only possible when the large government exists. You could argue that water…

Standard oil? Microsoft? At a high level government and big business are very similar in scope. It's not a matter of one or the other - your analysis is too simple. Government created the baby bells, not creative destruction. Government circumscribed ms.

Standard Oil was never a monopoly, and even its 90% market share had already dropped to 70% two years later when the government even filed charged, and even more when the case was decided.

Microsoft's market position is inseparable from its support by government in the form of software patents. When their rent on Android phones is larger than their income from their own phones, and that rent is purely based on having a patent on an obsolete filesystem which is only useful because it was previously "the standard" (ie., imposed by them), the government is directly helping Microsoft take advantage from their dominant market position in other markets - exactly what they purpose to ban.

Re: How Amazon Triggered a Robot Arms Race

#89

Post by kuschku »

antisthenes wrote: post #84
kuschku wrote: post #25

Earlier quoted context omitted.

And this is horrible for the market. A company gaining a huge advantage by getting a total monopoly over another market, which they can enforce due to having patents. That’s not what patents were supposed to create, and what Antitrust regulations were supposed to prevent. If this is the future, it’s quite a dystopia: Monopolies everywhere. Sure, I get that the Silicon Valley business model is to destroy regulation, g…

Properly regulated monopolies are the most efficient form of a market. In fact it is quite the opposite of what you're imagining.

I agree.

That’s why I usually consider government monopolies as better than for profit companies, as better than private monopolies.

(Considering the correlation between government monopolies being usually regulated better).

Sadly, in the past decades, we haven’t seen those in many places anymore.

The UK has mismanaged the NHS for many years, and is now considering to privatize it into a private monopoly.

Look at Comcast, TWC, etc, who have regional monopolies, and how well they work.

And considering how easy politicians are to bribe, I am not sure we can get properly regulated monopolies at all anymore.

Until we can ensure the proper regulation, we might want to prevent monopolies from even occuring.

Re: How Amazon Triggered a Robot Arms Race

#90

Post by Animats »

chillacy wrote: post #33
Animats wrote: post #18

I'd wondered how Kiva's old customers were doing. Now we know. There really isn't a direct replacement for Kiva robots from another vendor. The great thing about Kiva robots is that they're quite simple and robust mechanically. They're a battery, motors, wheels, and a lift mechanism. They're completely interchangeable, and having one fail isn't a big deal. Maintenance is replacing batteries, motors, wheels, and maybe…

These guys could be interesting, at least for smaller warehouses: http://www.inviarobotics.com/ Despite Kiva's simplicity, the warehouse has to be designed completely for robotic use, whereas these robots navigate around existing human accessible shelves. If Kiva is railway trains, these guys are automated cars: designed for the existing world.

Invia isn't shipping a product. They're just demoing a prototype under ideal conditions. Their robot doesn't look robust enough for a real warehouse. In a few years, they might be able to compete with Kiva, but not yet. If you have a distribution center with Kiva robots, and need to replace them now, you have a problem.

A Kiva warehouse is just a big space with a flat floor. It's not like the old Automated Storage and Retrieval Systems, with their custom tracks, conveyors, and towers.[1][2]

[1] https://www.youtube.com/watch?v=It7qr6fga3c [2] https://www.youtube.com/watch?v=LuvrP1QKsdk

Post reply on HN