US 12,242,931 B2
Systems and methods for soft model assertions
Daniel D. Kang, Fairfax, VA (US); Nikos Arechiga Gonzalez, Los Altos, CA (US); and Sudeep Pillai, Santa Clara, CA (US)
Assigned to TOYOTA RESEARCH INSTITUTE, INC., Los Altos, CA (US)
Filed by TOYOTA RESEARCH INSTITUTE, INC., Los Altos, CA (US)
Filed on May 10, 2021, as Appl. No. 17/316,551.
Prior Publication US 2022/0358401 A1, Nov. 10, 2022
Int. Cl. B60W 60/00 (2020.01); G06F 11/07 (2006.01); G06N 20/00 (2019.01)
CPC G06N 20/00 (2019.01) [B60W 60/001 (2020.02); G06F 11/0706 (2013.01); G06F 11/0751 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A system comprising:
a model assertion system monitoring one or more Machine Learning (ML) models to generate a severity score for each of the one or more ML models to indicate a type of error in a prediction of the ML model; and
a controller device receiving predictions generated based on the one or more ML models and the severity scores output from the model assertion system, and determining whether to perform autonomous operations of the system based on the severity score for each of the one or more ML models, wherein the autonomous operations are performed in response to the severity scores that indicate a type of error above a threshold level of certainty and comprise autonomously executing an evasive maneuver of a vehicle that corresponds to the severity score of each of the one or more ML models.