US 12,230,009 B2
Determining object behavior and trajectories
Hariprasad Govardhanam, Fremont, CA (US)
Assigned to GM Cruise Holdings LLC, San Francisco, CA (US)
Filed by GM Cruise Holdings LLC, San Francisco, CA (US)
Filed on Dec. 22, 2021, as Appl. No. 17/559,832.
Prior Publication US 2023/0196727 A1, Jun. 22, 2023
Int. Cl. G06T 7/20 (2017.01); G06V 10/74 (2022.01); G06V 10/762 (2022.01); G06V 10/82 (2022.01); G06V 20/50 (2022.01)
CPC G06V 10/762 (2022.01) [G06T 7/20 (2013.01); G06V 10/761 (2022.01); G06V 10/82 (2022.01); G06V 20/50 (2022.01); G06T 2207/30241 (2013.01)] 12 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
receiving sensor data corresponding with a trajectory of an object through an environment;
providing the sensor data to an autoencoder neural network to generate a first feature vector representing one or more behavioral attributes of the object in the environment, wherein the sensor data is provided to the autoencoder neural network to generate the first feature vector based on a threshold that limits sensor data provided to the autoencoder neural network to sensor data corresponding to objects whose predicted trajectories differ from trajectories that actually occurred for those objects;
classifying the one or more behavioral attributes based on the first feature vector, wherein the classifying the one or more behavioral attributes based on the first feature vector further comprises:
comparing the first feature vector to one or more pre-existing behavior clusters, and
classifying the one or more behavioral attributes based on a calculated similarity to at least one of the one or more pre-existing behavior clusters; and
determining one or more characteristics of a particular city based on a classification for feature embedding vectors of respective objects in the particular city.