US 11,911,901 B2
Training artificial networks for robotic picking
Yan Duan, Berkeley, CA (US); Haoran Tang, Emeryville, CA (US); Yide Shentu, Berkeley, CA (US); Nikhil Mishra, Irvine, CA (US); and Xi Chen, Berkeley, CA (US)
Assigned to Embodied Intelligence, Inc., Emeryville, CA (US)
Filed by Embodied Intelligence, Inc., Berkeley, CA (US)
Filed on Sep. 8, 2020, as Appl. No. 17/014,558.
Claims priority of provisional application 62/897,287, filed on Sep. 7, 2019.
Claims priority of provisional application 62/897,282, filed on Sep. 7, 2019.
Prior Publication US 2021/0069898 A1, Mar. 11, 2021
Int. Cl. B25J 9/16 (2006.01); B25J 15/06 (2006.01)
CPC B25J 9/161 (2013.01) [B25J 9/163 (2013.01); B25J 9/1612 (2013.01); B25J 9/1653 (2013.01); B25J 15/0658 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for training an artificial neural network, the method comprising:
performing, with a robotic picking device, an attempt to pick up an item at a first region on the item, wherein the first region corresponds to a first group of pixels;
in response to determining that the attempt failed, generalizing information about the first region, wherein the generalized information about the first region includes information about characteristics of the first region and an associated probability of producing the failed attempt based on the characteristics, and wherein the characteristics comprise locational features of the item at the first region with respect to the robotic picking device, one or more other regions on the item, and one or more other items;
transferring the generalized information about the first region to one or more other regions on the item, wherein transferring the generalized information comprises indicating a probability of failure associated with at least one different group of pixels having one or more of the same characteristics as the first group of pixels;
identifying a second region on the item to attempt to pick up the item from, wherein the second region is distinct from the first region and the one or more other regions, and wherein the second region corresponds to a second group of pixels lacking at least one characteristic of the first group of pixels;
performing, with the robotic picking device, a second attempt to pick up the item at the second region on the item;
in response to determining that the second attempt to pick up the item was unsuccessful, adjusting a probability of failure associated with the second region, identifying a perturbation strategy for repositioning the item based on the characteristics, and performing the perturbation strategy with a perturbation element of the robotic picking device; and
performing, with the robotic picking device, a third attempt to pick up the item at either the first region or the second region on the item, the first region or the second region selectively chosen based on the probabilities associated with the first region and the second region in a position following perturbation.