I had designed something like this with QR codes for an office. A camera would sit over top the desks to keep track of each folder and any stack that built up on top of it. That way anyone could quickly find what they need on any desk, instead of asking around.
Then, when storing and pulling any folder, it could go in any cabinet or box, instead of a previously defined, rigid, alphabetized and categorized location which would inevitably need to squeeze more than it could fit or waste a lot of space. A camera would (theoretically) automatically judge distance and location of a drawer in a storage room to accurately tell which one it went in or came from, or manually could be shown to the camera, as would need to be done in the case of a box. By consistently putting the folder in the front of the drawer or box, the computer could tell users the order and where exactly it sits (eg 2nd from the back, etc), which the color coded label would help identify.
QR codes still seem a little bit nicer than RFID here, except that Werkbank's image recognition goes even further than generic folders and makes good use of book cover images. It just seems like that robot costs a lot and has an unfortunate delay from when someone puts something away to when it updates a shelf's index. RFID seems better suited when it gets embedded in something like a toy or a farm animal, which a sensor can pick up near various "hot-spots", than when pasted flat on a book, tag, etc.
I'm so glad to see someone did this, and this library's (shared) approach seems to like a big step into Bret Victor's concept of Seeing Spaces: http://worrydream.com/SeeingSpaces/
I had designed something like this with QR codes for an office. A camera would sit over top the desks to keep track of each folder and any stack that built up on top of it. That way anyone could quickly find what they need on any desk, instead of asking around.
Then, when storing and pulling any folder, it could go in any cabinet or box, instead of a previously defined, rigid, alphabetized and categorized location which would inevitably need to squeeze more than it could fit or waste a lot of space. A camera would (theoretically) automatically judge distance and location of a drawer in a storage room to accurately tell which one it went in or came from, or manually could be shown to the camera, as would need to be done in the case of a box. By consistently putting the folder in the front of the drawer or box, the computer could tell users the order and where exactly it sits (eg 2nd from the back, etc), which the color coded label would help identify.
QR codes still seem a little bit nicer than RFID here, except that Werkbank's image recognition goes even further than generic folders and makes good use of book cover images. It just seems like that robot costs a lot and has an unfortunate delay from when someone puts something away to when it updates a shelf's index. RFID seems better suited when it gets embedded in something like a toy or a farm animal, which a sensor can pick up near various "hot-spots", than when pasted flat on a book, tag, etc.
I'm so glad to see someone did this, and this library's (shared) approach seems to like a big step into Bret Victor's concept of Seeing Spaces: http://worrydream.com/SeeingSpaces/
Found thanks to Sidnicious in another thread:
https://news.ycombinator.com/item?id=11266124
Ha, this reminds me of Mr. Penumbra's 24-Hour Bookstore. It's a good (nerdy) read!