US 12,403,596 B2
Imaging process for detecting failures modes
Yide Shentu, Berkeley, CA (US); David Mascharka, Emeryville, CA (US); Tianhao Zhang, Emeryville, CA (US); Yan Duan, Emeryville, CA (US); Jasmine Deng, Emeryville, CA (US); and Xi Chen, Emeryville, CA (US)
Assigned to Embodied Intelligence Inc., Emeryville, CA (US)
Filed by Embodied Intelligence Inc., Emeryville, CA (US)
Filed on Jul. 3, 2024, as Appl. No. 18/763,317.
Application 18/763,317 is a continuation of application No. 17/193,875, filed on Mar. 5, 2021, granted, now 12,053,887.
Claims priority of provisional application 62/985,982, filed on Mar. 6, 2020.
Prior Publication US 2024/0351203 A1, Oct. 24, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. B25J 9/16 (2006.01); G06N 3/02 (2006.01)
CPC B25J 9/1653 (2013.01) [B25J 9/1669 (2013.01); B25J 9/1687 (2013.01); B25J 9/1697 (2013.01); G06N 3/02 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method of detecting failure modes in a robotic motion environment, the method comprising:
collecting one or more images of a scene, wherein the one or more images are taken while a robot is in a first position of a motion cycle of the robot;
providing the one or more images of the scene taken while the robot is in the first position as a first input to a convolutional neural network;
determining, based at least in part on an output of the convolutional neural network, that a failure mode is not present in the scene, wherein the output is generated by the convolutional neural network based on the one or more images taken while the robot is in the first position;
collecting one or more new images of the scene, wherein the one or more new images are taken while the robot is in a second position, and wherein the second position occurs after the first position in the motion cycle of the robot;
providing the one or more new images of the scene taken while the robot is in the second position as a second input to the convolutional neural network;
determining, based at least in part on a second output of the convolutional neural network, that a second failure mode is present in the scene, wherein the second output is generated by the convolutional neural network based on the one or more new images taken while the robot is in the second position; and
in response to determining that the second failure mode is present in the scene, pausing motion of the robot.