US 11,938,939 B1
Determining current state of traffic light(s) for use in controlling an autonomous vehicle
Yan Li, San Francisco, CA (US); and Davis Edward King, Billerica, MA (US)
Assigned to AURORA OPERATIONS, INC., Mountain View, CA (US)
Filed by Aurora Innovation, Inc., Palo Alto, CA (US)
Filed on Sep. 30, 2020, as Appl. No. 17/039,517.
Claims priority of provisional application 63/034,522, filed on Jun. 4, 2020.
Int. Cl. G06N 20/00 (2019.01); B60W 30/18 (2012.01); B60W 60/00 (2020.01); G06V 20/58 (2022.01); G06V 20/56 (2022.01)
CPC B60W 30/18154 (2013.01) [B60W 60/001 (2020.02); G06N 20/00 (2019.01); B60W 2554/802 (2020.02); G06V 20/56 (2022.01); G06V 20/58 (2022.01); G06V 20/582 (2022.01); G06V 20/584 (2022.01)] 19 Claims
OG exemplary drawing
 
1. A method implemented by one or more processors of an autonomous vehicle, the method comprising:
processing an image using a machine learning classifier, wherein the image is captured by a vision component of the autonomous vehicle, and wherein the image is processed using the machine learning classifier based on the image capturing a region that is predicted to include a traffic light;
generating, based on processing the image using the machine learning classifier, a corresponding numerical measure for each of multiple candidate states of the traffic light;
generating, based on processing each of the corresponding numerical measures, a corresponding final measure for each of the candidate states;
processing an additional image using the machine learning classifier, wherein the additional image is captured by an additional vision component of the autonomous vehicle, wherein the additional image is processed using the machine learning classifier based on the additional image capturing the region that is predicted to include the traffic light, wherein the additional image is in addition to the image, wherein the additional image is captured at or near the same time as the image, and wherein the additional vision component is in addition to the vision component;
generating, based on processing the additional image using the machine learning classifier, an additional corresponding numerical measure for each of multiple candidate states of the traffic light;
generating, based on processing each of the additional corresponding numerical measures, an additional corresponding final measure for each of the candidate states;
selecting one of the candidate states as a current state of the traffic light, wherein the selecting is based on the corresponding final measures for the candidate states and based on the additional corresponding final measures for the candidate states; and
controlling the autonomous vehicle based on the selected current state of the traffic light.