US 11,934,995 B1
Package similarity prediction system
Huinan Ren, Seattle, WA (US); Nicole Yoshino, Seattle, WA (US); Kristin Quel Lien, Kirkland, WA (US); Abhishek Shrivastava, Bothell, WA (US); Vinit Dinesh Jain, Seattle, WA (US); Brian Gleadle, Kirkland, WA (US); Ruslan Khmeliuk, Bothell, WA (US); Michael Allen Swift, Brentwood, TN (US); and Pushkar Naik, Seattle, WA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Mar. 28, 2022, as Appl. No. 17/706,198.
Int. Cl. G06Q 10/087 (2023.01)
CPC G06Q 10/087 (2013.01) 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving, by a computer system, images of a package from a plurality of cameras associated with the computer system, the plurality of cameras located at a receiving area of a facility, the facility associated with the computer system;
generating, by the computer system, a visual vector for a vector space of package attributes using the images;
generating, by the computer system, a text vector for the vector space by executing an OCR algorithm that uses the images to extract text from the images;
receiving, by the computer system, physical dimensions and a weight of the package obtained by a plurality of sensors associated with the computer system;
generating, by the computer system, a combined vector that includes the visual vector and the text vector for the package;
determining, by the computer system, a subset of candidate packages of a plurality of historically received packages by comparing the physical dimensions and the weight of the package to the physical dimensions and the weight of the plurality of historically received packages;
determining, by the computer system, a ranking of the subset of candidate packages by identifying a distance in the vector space of the package attributes between a vector for each candidate package of the subset of candidate packages and the combined vector for the package using a nearest neighbor algorithm, the ranking of the subset of candidate packages representing a similarity between said each candidate package of the subset of candidate packages and the package within the vector space of the package attributes;
obtaining, by the computer system and for a particular package of the ranked subset of candidate packages, an identifier associated with the particular package;
retrieving, by the computer system and from a database, data for the particular package using the identifier, the database maintaining a mapping of identifiers to data for the plurality of historically received packages, the data including at least one of a quantity of an item for the particular package or a product identifier for the particular package;
generating, by the computer system, instructions for components of the facility to route the package from the receiving area of the facility to a particular location of the facility based at least in part on the data for the particular package; and
transmitting, by the computer system, the instructions to the components of the facility to transport the package from the receiving area of the facility to the particular location of the facility.