US 11,864,494 B2
AI-optimized harvester configured to maximize yield and minimize impurities
Dongyan Wang, San Jose, CA (US); Andrew Yan-Tak Ng, Los Altos, CA (US); Yiwen Rong, San Mateo, CA (US); Greg Frederick Diamos, Menlo Park, CA (US); Bo Tan, Santa Clara, CA (US); Beom Sik Kim, Belmont, CA (US); Timothy Viatcheslavovich Rosenflanz, Menlo Park, CA (US); Kai Yang, Fremont, CA (US); and Tian Wu, San Mateo, CA (US)
Assigned to Landing AI, Ugland House (KY)
Filed by Landing AI, Ugland House (KY)
Filed on Dec. 12, 2019, as Appl. No. 16/712,911.
Claims priority of provisional application 62/927,512, filed on Oct. 29, 2019.
Prior Publication US 2021/0120736 A1, Apr. 29, 2021
Int. Cl. G06N 20/00 (2019.01); A01B 79/00 (2006.01); G06T 7/00 (2017.01); G05B 13/02 (2006.01); G06N 5/04 (2023.01); G06V 20/10 (2022.01); A01D 41/127 (2006.01); H04N 23/90 (2023.01)
CPC A01D 41/127 (2013.01) [A01B 79/005 (2013.01); A01D 41/1274 (2013.01); G05B 13/0265 (2013.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01); G06T 7/0002 (2013.01); G06T 7/0004 (2013.01); G06T 7/97 (2017.01); G06V 20/188 (2022.01); H04N 23/90 (2023.01); G06T 2207/20081 (2013.01); G06T 2207/30188 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A computer-implemented method for detecting impurities of harvested plants in a receptacle of a harvester, the method comprising:
receiving, from a camera facing contents of the receptacle, an image of the contents;
applying the image as input to a first machine learning model;
receiving, as output from the first machine learning model, identification of an impurity of the harvested plants;
determining a current state of one or more physical components of the harvester;
applying, as input to a second machine learning model, both the current state of the one or more physical components of the harvester and the output from the first machine learning model of the identification of the impurity;
generating a control signal based on output of the second machine learning model;
transmitting the control signal to an operator interface that, responsive to receiving the control signal, provides an operator of the harvester with a recommended state change for the harvester;
determining a response of the operator to the recommended state change; and
causing the first machine learning model to adjust its outputs based on the response.