4 ms·

But how does it work? The article only mentions that you scan your phone at the entrance. The tags don't seem to be RFID or anything special neither.

jackhalford 6y ago
a bit of googling found a better article thats shows the cameras and sensors on the ceiling do the work. https://www.google.fr/amp/s/www.bbc.com/news/amp/technology-56266494 https://www.google.fr/amp/s/www.bbc.com/news/amp/technology-...
raspyberr 6y ago
Is that a critique of the article or are you unable to find that out online?
- jackhalford 6y ago
  I just found it, see sibling. Not a critique, honest curiosity.
devindotcom 6y ago
I wrote about it back when they showed off the first one here in Seattle. Not sure how much they've updated it but it's largely RGB cameras in the ceiling and weight sensors in the shelves: https://techcrunch.com/2018/01/21/inside-amazons-surveillance-powered-no-checkout-convenience-store/ https://techcrunch.com/2018/01/21/inside-amazons-surveillanc...
- londons_explore 6y ago
  Closer inspection of the shelves in the London store shows regular cheap steel shelving units with no weight sensors visible anywhere. Either the weight sensors are very well hidden, or there aren't any.
tuna-piano 6y ago
Amazon claims cameras automatically track what you take and put back. But if that's the case, then I'm left confused as to why receipts take so long (15-20 mins+ sometimes) to be generated after leaving the stores? Is it possible it's actually a manual process of people watching the videos, or more likely identifying what happened at each "take off shelf/put on shelf" point? They're still just building up the customer base, training data and models and it's not as magical as it all seems?
- londons_explore 6y ago
  They need to wait till not only you leave the ship, but also all people who were near you when you picked or returned any item. Then they can use that to resolve ambiguities ("one of these 3 people took this item which is now missing from the shelf, who was it")
- plumeria 6y ago
  Is your observation based on multiple samples? Or just a couple of casual visits? Would be interesting to know their real median receipt generation time.
plumeria 6y ago
Their patent [1] describes how it works. A nice summary is provided in [2]. [1] https://patents.google.com/patent/US10001402B1/en https://patents.google.com/patent/US10001402B1/en [2] https://www.geekwire.com/2018/fresh-patents-served-smart-shelf-technologies-seen-amazon-go-stores/ https://www.geekwire.com/2018/fresh-patents-served-smart-she...

jackhalford 6y ago

a bit of googling found a better article thats shows the cameras and sensors on the ceiling do the work. https://www.google.fr/amp/s/www.bbc.com/news/amp/technology-56266494 https://www.google.fr/amp/s/www.bbc.com/news/amp/technology-...

raspyberr 6y ago

Is that a critique of the article or are you unable to find that out online?

jackhalford 6y ago
I just found it, see sibling. Not a critique, honest curiosity.

devindotcom 6y ago

I wrote about it back when they showed off the first one here in Seattle. Not sure how much they've updated it but it's largely RGB cameras in the ceiling and weight sensors in the shelves: https://techcrunch.com/2018/01/21/inside-amazons-surveillance-powered-no-checkout-convenience-store/ https://techcrunch.com/2018/01/21/inside-amazons-surveillanc...

londons_explore 6y ago
Closer inspection of the shelves in the London store shows regular cheap steel shelving units with no weight sensors visible anywhere. Either the weight sensors are very well hidden, or there aren't any.

tuna-piano 6y ago

Amazon claims cameras automatically track what you take and put back. But if that's the case, then I'm left confused as to why receipts take so long (15-20 mins+ sometimes) to be generated after leaving the stores? Is it possible it's actually a manual process of people watching the videos, or more likely identifying what happened at each "take off shelf/put on shelf" point? They're still just building up the customer base, training data and models and it's not as magical as it all seems?

londons_explore 6y ago
They need to wait till not only you leave the ship, but also all people who were near you when you picked or returned any item. Then they can use that to resolve ambiguities ("one of these 3 people took this item which is now missing from the shelf, who was it")
plumeria 6y ago
Is your observation based on multiple samples? Or just a couple of casual visits? Would be interesting to know their real median receipt generation time.

plumeria 6y ago

Their patent [1] describes how it works. A nice summary is provided in [2]. [1] https://patents.google.com/patent/US10001402B1/en https://patents.google.com/patent/US10001402B1/en [2] https://www.geekwire.com/2018/fresh-patents-served-smart-shelf-technologies-seen-amazon-go-stores/ https://www.geekwire.com/2018/fresh-patents-served-smart-she...