US 12,455,358 B1
Machine-learned classifier for lidar split returns
Samantha Marie Ting, Redwood City, CA (US); Venkata Subrahmanyam Chandra Sekhar Chebiyyam, San Francisco, CA (US); and Shaminda Subasingha, San Ramon, CA (US)
Assigned to Zoox, Inc., Foster City, CA (US)
Filed by Zoox, Inc., Foster City, CA (US)
Filed on Dec. 17, 2021, as Appl. No. 17/555,359.
Int. Cl. G01S 7/497 (2006.01); G01S 7/40 (2006.01)
CPC G01S 7/497 (2013.01) [G01S 7/4021 (2013.01); G01S 7/4039 (2021.05); G01S 2007/4975 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
receiving sensor data associated with an environment from a sensor;
determining that a return associated with the sensor data includes a first split return, wherein the first split return indicates two different surfaces associated with a same direction from the sensor;
determining a set of split returns including the first split return and an additional split return;
determining, by a machine-learned (ML) model, a first confidence score indicating that the first split return is associated with surfaces of two objects and a second confidence score indicating that the first split return is associated with particulate matter and a second surface; and
controlling a vehicle based at least in part on at least one of the first confidence score or the second confidence score.