US 12,475,580 B1
System for aligning sensor data with maps comprising covariances
Patrick Blaes, Redwood City, CA (US); Michael Carsten Bosse, Cupertino, CA (US); Elena Stephanie Stumm, San Francisco, CA (US); Derek Adams, Pasadena, CA (US); and Brice Rebsamen, Aix-en-Provence (FR)
Assigned to Zoox, Inc., Foster City, CA (US)
Filed by Zoox, Inc., Foster City, CA (US)
Filed on Jan. 23, 2024, as Appl. No. 18/420,346.
Application 18/420,346 is a continuation of application No. 17/446,344, filed on Aug. 30, 2021, granted, now 11,915,436.
Int. Cl. G06T 7/33 (2017.01); G06T 3/20 (2006.01); G06T 3/60 (2006.01)
CPC G06T 7/337 (2017.01) [G06T 3/20 (2013.01); G06T 3/60 (2013.01); G06T 2207/20221 (2013.01); G06T 2207/20224 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method comprising:
receiving first data representative of a physical environment;
receiving second data representative of the physical environment;
aligning the first data with the second data to generate map data of the physical environment as a process that iterates until a resolution threshold is met or exceed, a subsequent resolution of an individual iteration being performed at a finer resolution than a previous resolution, individual iterations of the process comprising:
associating a portion of the first data with a first discretized region at a current resolution;
determining, based at least in part on a subset of the first data associated with a first subregion of the first discretized region, a first covariance;
associating a portion of the second data with a second discretized region at the current resolution;
determining, based at least in part on a subset of the second data associated with a second subregion of the second discretized region, a second covariance;
determining, based at least in part on the first covariance and the second covariance, a transformation indicating one or more of a rotation or a translation between the first data and the second data;
determining a weighted sum of a Gaussian distribution of the first discretized region and a Gaussian distribution of the second discretized region;
generating, as the map data, a combined covariance based at least in part on the first covariance, the second covariance, and the weighted sum; and
outputting the map data.