US 11,741,692 B1
Prediction error scenario mining for machine learning models
Juraj Kabzan, Boston, MA (US); Sammy Omari, Pittsburgh, PA (US); and Julia Gomes, San Francisco, CA (US)
Assigned to Motional AD LLC, Boston, MA (US)
Filed by Motional AD LLC, Boston, MA (US)
Filed on Dec. 9, 2022, as Appl. No. 18/78,594.
Application 18/078,594 is a continuation of application No. 17/673,633, filed on Feb. 16, 2022, granted, now 11,562,556.
Int. Cl. G06V 20/58 (2022.01); G06V 10/774 (2022.01); G06T 7/20 (2017.01)
CPC G06V 10/7747 (2022.01) [G06T 7/20 (2013.01); G06V 20/58 (2022.01); G06T 2207/10028 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/30261 (2013.01)] 30 Claims
OG exemplary drawing
 
1. A system, comprising:
at least one data processor; and
at least one memory storing instructions, which when executed by at least one data processor, result in operations comprising:
identifying a first scenario encountered by a vehicle, the first scenario being identified based at least on a firsterror in an output of an online machine learning model trained to plan a movement of the vehicle while the vehicle encounters the first scenario satisfying a first threshold;
generating a first label for the first scenario by applying an offline machine learning model also trained to plan the movement of the vehicle;
generating training data including the first scenario and the first label associated with the first scenario; and
updating, based at least on the training data, the online machine learning model to plan the movement of the vehicle.