US 11,941,572 B1
Machine learning item delivery using sensor data
Ruth Ravichandran, Redmond, WA (US); Hebaallah Aly Abdelhalim Aly Ismail, Bellevue, WA (US); Zheng Wang, Sammamish, WA (US); Shao-Wen Yang, New York, NY (US); Yang Pan, San Jose, CA (US); David Hung Huynh, San Jose, CA (US); Andrey Li, San Jose, CA (US); Hoshgeldy Tagangeldyevich Tachmuradov, Soquelq, CA (US); and Steven Larson, Georgetown, TX (US)
Assigned to AMAZON TECHNOLOGIES, INC., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Mar. 16, 2021, as Appl. No. 17/203,098.
Int. Cl. G06Q 10/0833 (2023.01); G06N 7/01 (2023.01); G06N 20/00 (2019.01); G06T 7/579 (2017.01); G06T 19/00 (2011.01); H04W 4/02 (2018.01); H04W 4/024 (2018.01); H04W 4/38 (2018.01); H04W 4/80 (2018.01)
CPC G06Q 10/0833 (2013.01) [G06N 7/01 (2023.01); G06N 20/00 (2019.01); G06T 7/579 (2017.01); G06T 19/006 (2013.01); H04W 4/023 (2013.01); H04W 4/024 (2018.02); H04W 4/026 (2013.01); H04W 4/027 (2013.01); H04W 4/38 (2018.02); H04W 4/80 (2018.02)] 20 Claims
OG exemplary drawing
 
1. A non-transitory computer-readable medium having program instructions stored thereon executable by at least one hardware processor that, when executed, direct the at least one hardware processor to:
receive peripheral device data from at least one peripheral device associated with a delivery driver, the peripheral device comprising at least one sensor configured to generate the peripheral device data;
determine, through execution of a machine learning routine, a current activity and a current location of the delivery driver relative to a delivery destination;
in response to the delivery driver being within a threshold distance of the delivery destination, determine a current state of a delivery process through execution of a machine learning routine using the peripheral device data as an input to the machine learning routine;
transition a client application executing on an augmented reality device associated with the delivery driver between a plurality of states based at least in part on the current state of the delivery process, wherein the transition of the client application between states is performed without a manual interaction with the augmented reality device;
wherein at least one of the states of the client application comprises providing navigation directions in an augmented reality environment provided by the augmented reality device; and,
wherein:
in an instance in which the current activity indicates that the delivery driver is driving to a delivery destination, the machine learning routine utilizes a first machine learning model;
in an instance in which the current activity indicates that the delivery driver is seeking or retrieving the item from a cargo area of a vehicle, the machine learning routine utilizes a second machine learning model; and
in an instance in which the current activity indicates that the delivery driver is walking to or from a delivery destination to the vehicle, the machine learning routine utilizes a third machine learning model.