US 11,704,385 B2
Traffic light detection auto-labeling and federated learning based on vehicle-to-infrastructure communications
Kun-Hsin Chen, Mountain View, CA (US); Sudeep Pillai, Santa Clara, CA (US); Shunsho Kaku, Mountain View, CA (US); Hai Jin, Ann Arbor, MI (US); Peiyan Gong, Ann Arbor, MI (US); and Wolfram Burgard, Mountain View, CA (US)
Assigned to TOYOTA RESEARCH INSTITUTE, INC., Los Altos, CA (US)
Filed by TOYOTA RESEARCH INSTITUTE, INC., Los Altos, CA (US)
Filed on Sep. 25, 2020, as Appl. No. 17/33,169.
Prior Publication US 2022/0101045 A1, Mar. 31, 2022
Int. Cl. G06K 9/00 (2022.01); G06F 18/214 (2023.01); B60W 60/00 (2020.01); G06V 20/58 (2022.01); G06F 18/21 (2023.01)
CPC G06F 18/214 (2023.01) [B60W 60/001 (2020.02); G06F 18/217 (2023.01); G06V 20/584 (2022.01); B60W 2552/00 (2020.02); B60W 2555/60 (2020.02); B60W 2556/45 (2020.02)] 18 Claims
OG exemplary drawing
 
1. A method for traffic light auto-labeling, comprising:
aggregating vehicle-to-infrastructure (V2I) traffic light signals at an intersection to determine transition states of each driving lane at the intersection during operation of an ego vehicle;
automatically labeling image training data to form auto-labeled image training data for a traffic light recognition model within the ego vehicle according to the determined transition states of each driving lane at the intersection;
planning a trajectory of the ego vehicle to comply with a right-of-way according to the determined transition states of each driving lane at the intersection according to a trained traffic light detection model;
replacing a current version of the traffic light recognition model during operation of the ego vehicle with an updated version of the traffic light recognition model when a performance of the updated version of the traffic light recognition model is greater than a predetermined value; and
operating the ego vehicle according to the updated version of the traffic light recognition model.