Comment on Amazon made a new version of its cashierless tech that doesn’t need cameras

geosoco@kbin.social ⁨1⁩ ⁨year⁩ ago

• Noumena@kbin.social ⁨1⁩ ⁨year⁩ ago

  I see teo reported major issues

  First, the made the product manufacturers include them on their own dime. I get the logic, have that manufacturer include adding them in their assembly line process. But as stated, this was an extra expense for a couple of reasons and at times made the product unprofitable.

  The solution, to start, seems to be either adding them via a walmart processing center, or have a funds process where sellers could get walmart to refund the cost and share it. I guess the third option would be to raise all prices at walmart to ensure the cost was bulit into the product.

  Second problem, data load. Compared to 2006, i think we are much better at large datsets these days both from a space and processing power perspective. Datacenters in 2006 were often on-prem with upgrades to size having large and expensive lead times. AWS changed that and Amazon owns that, so i expect that bottle neck to be solvable.

  source

  • geosoco@kbin.social ⁨1⁩ ⁨year⁩ ago

    There's at least one additional issue, and I think it's something Walmart ran into when trying the RFID for checkout and it's the noisy radio environment which led to issues scanning all of the codes properly or including other people's items as one of your purchases if it's too close (eg. the self checkout counters being close by or shopping with a friend/partner who is behind you.)

    source

    • Noumena@kbin.social ⁨1⁩ ⁨year⁩ ago

      Yea, that does sound like breaking problem. I have no clue how to begin solving that one.

      I feel like there was an idea, not in this article, where people would scan items as they add them to their cart and then this walkout process was really just a redundant verification of walking out with the purchase. Still, that does sound like a lot of error prone noise.

      source

      • geosoco@kbin.social ⁨1⁩ ⁨year⁩ ago

        I don't know what's been tried in a real retail market, but I remember Walmart or perhaps just designers trying all sorts of ideas from carts that scanned them to people using their phones. It starts to get really complicated when people want to remove things or you have multiple people putting stuff in carts (eg. children sneaking items in)

        I'm sure it's solvable, it's just a question of whether it's worth the cost for shoppers and the stores.

        source