	US 12,189,718 B1
	Learned state covariances
Subhasis Das, Menlo Park, CA (US); and Shida Shen, San Mateo, CA (US)
Assigned to Zoox, Inc., Foster City, CA (US)
Filed by Zoox, Inc., Foster City, CA (US)
Filed on Dec. 22, 2022, as Appl. No. 18/087,374.
Application 18/087,374 is a continuation of application No. 16/862,911, filed on Apr. 30, 2020, granted, now 11,537,819.
Int. Cl. G06V 20/56 (2022.01); G06F 18/21 (2023.01); G06F 18/24 (2023.01); G06F 18/25 (2023.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01); G06T 7/277 (2017.01); H04W 4/44 (2018.01)

CPC G06F 18/217 (2023.01) [G06F 18/24 (2023.01); G06F 18/25 (2023.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01); G06T 7/277 (2017.01); G06V 20/56 (2022.01); H04W 4/44 (2018.02); G06T 2207/10028 (2013.01); G06T 2207/10044 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30252 (2013.01)]

21 Claims

1. A method comprising:

receiving sensor data;

determining observation data associated with an object represented in the sensor data;

inputting the observation data associated with the object represented in the sensor data into a machine-learning (ML) model;

receiving, as output from an output head of a plurality of output heads of the ML model, an observation covariance value associated with the observation data associated with the object represented in the sensor data, wherein an individual output head of the plurality of output heads is associated with one or more of an object classification, an environment classification, and a sensor type, and

wherein the observation covariance value represents a measure of joint variability of two or more instances of the observation data;

determining, based at least in part on the observation covariance value, track data associated with the object represented in the sensor data; and

outputting the track data to a planning component of an autonomous vehicle.