	US 12,112,432 B2
	Realistic 3D virtual world creation and simulation for training automated driving systems
Dan Atsmon, Rehovot (IL)
Assigned to Cognata Ltd., Rehovot (IL)
Filed by Cognata Ltd., Rehovot (IL)
Filed on Aug. 11, 2022, as Appl. No. 17/885,633.
Application 17/885,633 is a continuation of application No. 16/693,534, filed on Nov. 25, 2019, granted, now 11,417,057.
Application 16/693,534 is a continuation of application No. 16/313,058, granted, now 10,489,972, issued on Nov. 26, 2019, previously published as PCT/IL2017/050598, filed on May 29, 2017.
Claims priority of provisional application 62/384,733, filed on Sep. 8, 2016.
Claims priority of provisional application 62/355,368, filed on Jun. 28, 2016.
Prior Publication US 2022/0383591 A1, Dec. 1, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06T 19/00 (2011.01); G01C 21/00 (2006.01); G05D 1/00 (2024.01); G06F 30/20 (2020.01); G06T 17/05 (2011.01)

CPC G06T 17/05 (2013.01) [G01C 21/3815 (2020.08); G01C 21/3826 (2020.08); G01C 21/3867 (2020.08); G05D 1/0088 (2013.01); G06F 30/20 (2020.01); G06T 19/006 (2013.01)]

20 Claims

1. A system for creating a simulated virtual realistic model of a geographical area for training, evaluating or validating an autonomous driving system (ADS), comprising:

at least one processor adapted to execute code, the code comprising:

code instructions to obtain geographic map data of a geographical area;

code instructions to obtain visual imagery data of the geographical area;

code instructions to classify a plurality of static objects identified in the visual imagery data to corresponding labels to designate a plurality of labeled objects;

code instructions to superimpose the plurality of labeled objects over the geographic map data;

code instructions to generate a virtual three dimensional (3D) realistic ADS training model emulating the geographical area by synthesizing a corresponding visual texture for each of the plurality of labeled objects; and

code instructions to create the simulated virtual realistic model by inserting one or more dynamic objects to the virtual 3D realistic ADS training model; and

code instructions to apply movement of said one or more dynamic objects according to movement patterns predefined or learned for said geographical area.