US 11,657,572 B2
Systems and methods for map generation based on ray-casting and semantic class images
Yong-Dian Jian, Campbell, CA (US)
Assigned to argo ai, llc, Pittsburgh, PA (US)
Filed by Argo AI, LLC, Pittsburgh, PA (US)
Filed on Oct. 21, 2020, as Appl. No. 17/75,827.
Prior Publication US 2022/0122324 A1, Apr. 21, 2022
Int. Cl. G06T 17/20 (2006.01); G06T 15/08 (2011.01); G06T 15/06 (2011.01)
CPC G06T 17/205 (2013.01) [G06T 15/06 (2013.01); G06T 15/08 (2013.01)] 21 Claims
OG exemplary drawing
 
1. A method for generating a map, comprising:
obtaining, by a computing device, a 3D point cloud combining a plurality of 3D point clouds that comprise sensor data generated by sensors and aligned using localization and mapping operations;
creating, by the computing device, a grid of voxels for the 3D point cloud, where each of said voxel comprises a single point of the 3D point cloud;
identifying, by the computing device, static occupied voxels and dynamic occupied voxels based on results of ray-casting operations, wherein each one of said static occupied voxels comprises a first data point of the 3D point cloud that has a single line of sight passing therethrough and each one of the dynamic occupied voxels comprises a second data point of the 3D point cloud that has at least two lines of sight passing therethrough, where a first one of the lines of sight extends from a first vantage point or location of a 3D laser scanner and a second one of the lines of sight extends from a different second vantage point or location of the 3D laser scanner;
generating, by the computing device, a pruned 3D point cloud by removing the second data points from the 3D point cloud that are associated with the dynamic occupied voxels;
generating, by the computing device, a 2D binary mask for at least one semantic label class of the pruned 3D point cloud;
processing, by the computing device, the 2D binary mask to obtain x-coordinates and y-coordinates for a 2D volume defining an object of the at least one semantic label class;
identifying, by the computing device, second data points in the pruned 3D point cloud based on the 2D volume;
comparing, by the computing device, z-coordinates of the second data points to at least one threshold value selected for the at least one semantic label class; and
generating, by the computing device, the map by removing one or more of the second data points from the pruned 3D point cloud based on results of the comparing.