US 12,228,936 B2
Method and system for deterministic trajectory selection based on uncertainty estimation for an autonomous agent
Gautam Narang, Palo Alto, CA (US); Apeksha Kumavat, Palo Alto, CA (US); Arjun Narang, Palo Alto, CA (US); Kinh Tieu, Palo Alto, CA (US); Michael Smart, Palo Alto, CA (US); and Marko Ilievski, Palo Alto, CA (US)
Assigned to Gatik AI Inc., Mountain View, CA (US)
Filed by Gatik AI Inc., Palo Alto, CA (US)
Filed on Jan. 13, 2023, as Appl. No. 18/096,695.
Application 18/096,695 is a continuation of application No. 17/704,301, filed on Mar. 25, 2022, granted, now 11,586,214.
Application 17/704,301 is a continuation of application No. 17/685,905, filed on Mar. 3, 2022, granted, now 11,487,296, issued on Nov. 1, 2022.
Application 17/685,905 is a continuation of application No. 17/496,053, filed on Oct. 7, 2021, granted, now 11,320,827, issued on May 3, 2022.
Application 17/496,053 is a continuation of application No. 17/483,539, filed on Sep. 23, 2021, granted, now 11,307,594, issued on Apr. 19, 2022.
Application 17/483,539 is a continuation of application No. 17/127,599, filed on Dec. 18, 2020, granted, now 11,157,010, issued on Oct. 26, 2021.
Claims priority of provisional application 63/055,794, filed on Jul. 23, 2020.
Claims priority of provisional application 63/035,401, filed on Jun. 5, 2020.
Prior Publication US 2023/0152816 A1, May 18, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G05D 1/00 (2024.01); G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06N 20/00 (2019.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 20/56 (2022.01)
CPC G05D 1/0219 (2013.01) [G05D 1/0088 (2013.01); G05D 1/0214 (2013.01); G05D 1/0221 (2013.01); G06F 18/2155 (2023.01); G06F 18/217 (2023.01); G06N 20/00 (2019.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 20/56 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A method for operating an autonomous vehicle, the method comprising:
determining a 1st trajectory for an autonomous vehicle, wherein the 1st trajectory is determined with a set of learned models;
operating the autonomous vehicle according to the 1st trajectory;
during operation of the autonomous vehicle according to the 1st trajectory:
detecting a failure associated with operation of the autonomous vehicle;
characterizing the failure with an out-of-distribution detector;
based on the characterized failure, retrieving a programmed trajectory;
operating the autonomous vehicle according to the programmed trajectory.