US 12,269,488 B2
End-to-end evaluation of perception systems for autonomous systems and applications
David Nister, Bellevue, WA (US); Cheng-Chieh Yang, Sunnyvale, CA (US); and Yue Wu, Mountain View, CA (US)
Assigned to NVIDIA Corporation, Santa Clara, CA (US)
Filed by NVIDIA Corporation, Santa Clara, CA (US)
Filed on Apr. 21, 2022, as Appl. No. 17/726,407.
Claims priority of provisional application 63/177,816, filed on Apr. 21, 2021.
Prior Publication US 2022/0340149 A1, Oct. 27, 2022
Int. Cl. B60W 50/02 (2012.01); B60W 60/00 (2020.01)
CPC B60W 50/0205 (2013.01) [B60W 60/001 (2020.02); B60W 2050/021 (2013.01); B60W 2420/403 (2013.01); B60W 2554/404 (2020.02)] 30 Claims
OG exemplary drawing
 
1. At least one processor comprising:
one or more circuits to:
receive ground truth data and perception data, the perception data generated based at least one or more machine learning models processing sensor data captured at a time corresponding to the ground truth data;
compute, based at least a first forward simulation, one or more first control decisions of one or more machine components processing first input data corresponding to the ground truth data;
compute, based at least on a second forward simulation, one or more second control decisions of the one or more machine components processing second input data corresponding to the perception data;
determine whether a fault is present based at least on comparing the one or more first control decisions to the one or more second control decisions; and
compute one or more key performance indicators (KPIs) based at least on the determination of whether the fault is present.