10 ms·
How Amazon Triggered a Robot Arms Race
- peterkshultz 10y agoAmazon's acquisition of Kiva is one of the best examples of vertical integration in the e-commerce industry. The article reminded me of acquisitions in the Gilded Age. Perhaps one could compare Bezos to Rockefeller, presuming Bezos was the one who pushed for such purchases.
- prawn 10y agoFor anyone yet to read the article who is initially confused (as I was), besides the Kiva you might know from microlending, there's also a Kiva who make industrial robots for warehouses. As per the article, Amazon bought the latter Kiva and forced competitors to turn to alternatives that were yet to be developed.
- NetTechM 10y agoThanks for that, I was initially "lost in the sauce" as it were.
- icebraining 10y agoBezos is no Rockefeller: https://en.wikipedia.org/wiki/John_D._Rockefeller#Philanthropy https://en.wikipedia.org/wiki/John_D._Rockefeller#Philanthro...
- Aelinsaar 10y agoInteresting. I wonder if this system and that "Hover" technology would be a good fit? A warehouse seems like one of the good places to line the floor with metal and clear the people out.
- jethro_tell 10y agoNot sure what hovering would do for these little guys but the good news about a robotic forklift running over a robotic inventory picker is you just file a bug report not an accident claim. At the least it seems like a great way to scale insurance, benefit, and labor management costs.
- taneq 10y agoYou mean the electromagnetic 'hoverboard' thing? This is probably the best bet for that tech to become relevant, but it's hard to see how it'd be more efficient than just using air cushions, which are an established way of moving heavy things around on smooth floors (let alone wheels).
- Aelinsaar 10y agoThat's what I was thinking of, and to be honest, don't know much about it. Maybe there's a potential maintenance advantage, given fewer moving parts?
- deegles 10y agoPersonally, I think it's a net benefit that the picking jobs are on their way out. I don't believe most people do it for fun! On the other hand, there's going to be short term turmoil as people scramble to find new jobs in an increasingly automated world.
- peterkshultz 10y agoI'd imagine that people who need such picking jobs would rather have them around than not.
- ethbro 10y agoIsn't that the point of UBI? Getting us past the local maxima of "this job is soul-crushing and a waste of a person's intelligence, but economics make it cheaper to use a body than a robot"?
- peterkshultz 10y agoFair. But a soul-crushing job is better than no job, and the question remains as to whether or not mobility into a job of equal requisite skill exists.
- timv 10y ago> a soul-crushing job is better than no job Really? Do you mean: "Being paid to do soul-crushing work is better than not being paid" ? Or: "Doing soul-crushing work is better than having nothing to do" ? The first is an issue of income and the distribution of product (in the GDP sense of product). There is no axiomatic rule that says "you have to have a job to get paid", it's just how our current economy operates. If I owned a farm that produced a variety of food in excess of the needs of my whole family, but was totally operated by self-maintaining robots, would I need to have a job? The robots would produce all the food I need, and I could sell the excess in order to obtain other goods. I guess I could say my job was "farmer" or "grocery salesman" or "robot supervisor", but I wouldn't actually being doing work. If I am a truck driver, and I own my own truck and I make an income by transporting goods, and then I convert my truck to be self-driving so that I no longer need to be the driver, then (if we temporarily ignore external price pressures) I still make the same income as I used to. But I'm just reaping the benefits of my self driving truck, I don't actually "work" anymore (or at least to nowhere near the extent I used to). But if I am a truck driver, and someone else owns the truck and I make an income by being contracted to transport goods, and then they convert my truck to be self-driving so that I no longer need to be the driver, then I no longer make any income, and I'm screwed. The owner of the truck reaps incredible benefits, and I don't actually "work" anymore, but I also don't have an income. That's how things work in our society, but it doesn't have to be that way. There's no universal moral law that says that the first ex-truck-driver should continue to be paid for doing nothing much, but the second ex-truck-driver needs to get off his butt and go find a new soul-crushing job so that he can earn an income. As our economies change to produce more output with fewer workers, there becomes less useful work for the citizens to do, even though GDP is increasing. That ought to be a good thing (economically) - more "stuff we want", less effort to produce it. But it won't actually being a good thing for the general population unless we move to a different model for the distribution of income, so that there's a way for the populace to share in the increased productivity even though their manual labour is no longer needed. That where ideas like "Universal Basic Income" (UBI) come in.
- ChuckMcM 10y agoI was reminded of how Dell originally pioneered extreme operational efficiency as the tool which made them hard to beat. I also particularly like the warehouse robotics application, it is a good mix of being able to make some adaptations in the environment to mitigate some of the more serious robotics challenges and yet able to effectively evaluate the net value to the business with and without. Just wish I'd started a warehouse robot company 4 years ago :-)
- iandanforth 10y agoNot too late to join one! We (Fetch Robotics) are hiring for many roles!
- desdiv 10y agoFYI the link on your profile, http://fetchrobotics.com/?page_id=30 http://fetchrobotics.com/?page_id=30, is bouncing to an outdated page: http://fetchrobotics.com/careers-old http://fetchrobotics.com/careers-old.
- iandanforth 10y agoFixed. Thanks!
- are595 10y agoAs a Computer Engineering student, what sort of courses do you recommend I take in order to work on robotics? My only experience with the field so far has been high school robotics competitions really.
- iandanforth 10y agoRobotics is a hugely interdisciplinary field. Some of the hardest parts to work on are motion planning and navigation. For us familiarity with ROS is pretty important and project courses where you build something that actually works are super important. Search algorithms come up all the time, spatial searching especially. For specifics I'd refer to the CMU Masters courseload or the WPI undergrad: https://www.ri.cmu.edu/ri_static_content.html?menu_id=469 https://www.ri.cmu.edu/ri_static_content.html?menu_id=469 http://www.wpi.edu/academics/robotics/ugprograms.html http://www.wpi.edu/academics/robotics/ugprograms.html
- Animats 10y agoI'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 the electronics module as necessary. The infrastructure they require is modest. The competing systems all seem to require more elaborate robots, special totes, and special shelf construction. Amazon says that Kiva robots cut their order picking time from 90 minutes to 15. Other retailers, seeing that number, must be terrified. It's a technological improvement like that which makes same-day delivery from a large inventory possible.
- kuschku 10y agoAnd 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, get a monopoly (ideally in a new market niche, but destroying existing competition with illegal predatory pricing is also common), and then to abuse the monopoly, but is no one thinking that for the average consumer a free, competitive market might be better?
- ZeroGravitas 10y agoThis is my theory as to why Europe can't compete with silicon valley. It turns out, what they're "competing" on is willingness to turn a blind eye to monopolies and their negative economic effects.
- zorked 10y agoAre these effects really negative then?
- jpeg_hero 10y agoI thought Amazon was pulling back on Kiva. I thought I read somewhere that new warehouses were not using the Kiva systems.
- codeonfire 10y agoThey can't be used for large item warehouses and are probably not worth it for smaller warehouses. Also existing warehouses that are multilevel with plywood flooring probably can't be retrofitted easily. Since their focus seems to be on same day, one hour delivery from small urban warehouses, the majority of their new warehouses are probably not going to use kiva.
- petra 10y agoI don't think 1-day will be that important, 2-day is fine for most orders. But the warehouses help them save a lot of money on shipping.But surely the saving on opex in a small robotic warehouse is less than a big robotic warehouse.
- huherto 10y agoWhy are they not worht for smaller warehouses ? The cost of kiva robots for amazone are per unit. Thus proportional to the size of the warehouse.
- jonknee 10y agoA robot that goes and fetches items and brings them to you is more valuable in larger spaces than smaller ones.
- codeonfire 10y agoCost of installing the system, cost of maintenance for a smaller number of bots, less dense shelving, more specialized training, lower picking costs because of smaller area. Prime now also delivers refrigerated food which probably Kiva can't do. Probably smaller warehouses are not purpose built for Kiva so maybe they don't have power, floor space, or layout to make kiva work well. This one in Manhattan is only 40k feet and has low ceilings: http://time.com/4159144/amazon-prime-warehouse-new-york-city-deliveries-christmas/ http://time.com/4159144/amazon-prime-warehouse-new-york-city...
- kristofferR 10y agoKomplett.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/ http://autostoresystem.com/ https://www.youtube.com/watch?v=_O0UKARdLDc https://www.youtube.com/watch?v=_O0UKARdLDc (360° video) http://www.tu.no/artikler/unik-video-bli-med-inn-i-kompletts-splitter-nye-robotlager/223905 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 would be easier to automate that for the AutoStore system since each product has its own box as opposed to Amazons Kiva system where multiple products are placed on the same shelf.
- sschueller 10y agoI find autostore to be a much cleaner solution than the adhoc system amazon uses. With the amazon system you need a lot more space and staff to fill those shelves. In addition you have to be careful not to load them top heavy.
- karambahh 10y agoThe only downside I see with Autstore is building maintenance. What about maintenance on a column? You have to stop the whole system, unload & disassemble part of the grid just to run a new cable or check for humidity or whatever on your structural columns. That's the only downside I see. It's a very simple concept solving a very real issue and their motto "stop airhousing start warehousing" real hits home!
- sschueller 10y agoFrom what it looks like it may not be so bad. Since the system appears to be dynamically scalable you can most likely move all items out of a section and make it off limits. Then do maintenance etc on that section.
- Shivetya 10y agolooking at autostore I was curious how many skus can be done efficiently and how deep you can go relative to the number of fast moving lines? With a million SKUs it might be fun to see what initial mapping can be done. Also, how can hazardous stacks be integrated to they can deliver to the same work station as regular items? as for maintenance I would think a few strategic extra rows could serve that purpose without losing too much efficiency. most the picking I am used to has both voice and light based picking with high velocity parts duplicated to the front.
- reachtarunhere 10y agoGreyOrange (http://www.greyorange.com http://www.greyorange.com) is a well funded startup solving the exact problem.
- mooreds 10y agoSuper exciting, because of the efficiencies, but I hear echoes of manna: http://marshallbrain.com/manna1.htm http://marshallbrain.com/manna1.htm
- justlsice 10y agoFrom what I know from a friend that works there, Kiva robots are still not profitable enough, even though they're quite simple tech-wise.
- huherto 10y agoThe innovation is great. But it is scary that amazon may start buying all these robot startups just to keep the patents and their products out of the hand of other e-commerce companies.
- sgwealti 10y agoThis article really overhypes the Kiva robots. I work in the warehouse automation industry for a fortune-500 retailer with several large e-commerce brands. Robotic goods-to-person is only one of many applications of robotics and automation in warehouses. It is also really only useful for certain types of long-tail operations - many SKUs, low volume per SKU. It also still very labor intensive because someone still has to pack all of those orders. I think the Kiva robots are an important part of Amazon's fulfillment efficiency but I don't see a huge market for that technology across the warehousing industry because there are a lot of other problems that it doesn't solve at all or doesn't handle very well. From my point of view the automated storage and retrieval and shuttle systems are much more interesting and more widely applicable. They are also a much more efficient use of vertical space.
- endersshadow 10y agoOne of the nice things about the Kiva robots is that they will actually organize the warehouse based on popularity of the SKUs automagically. Things that are low volume get placed farther away from the pickers, and things that are high volume are never far away from the pickers.
- sgwealti 10y agoI agree that is an advantage of the goods-to-person robots but shuttle systems do the same thing and are a better use of vertical space.
- endersshadow 10y agoWhat's an example of them? I'm interested to learn.
- sgwealti 10y agoThere 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 or more order lines of a particular SKU, the system can run into scaling issues. There are only so many robots and so much floor space to navigate in. Shuttles to some extent have the same problem but because of their smaller footprint you run out of horizontal space slower than you do with Kiva robots.
- SlipperySlope 10y agoTrump is right to go after Amazon on monopoly and antitrust.