US 12,460,940 B2
Machine learning based occupancy grid generation
Volodimir Slobodyanyuk, San Diego, CA (US); Radhika Dilip Gowaikar, San Diego, CA (US); Makesh Pravin John Wilson, San Diego, CA (US); Shantanu Chaisson Sanyal, San Diego, CA (US); Avdhut Joshi, San Marcos, CA (US); Christopher Brunner, San Diego, CA (US); Behnaz Rezaei, San Diego, CA (US); and Amin Ansari, Federal Way, WA (US)
Assigned to QUALCOMM Incorporated, San Diego, CA (US)
Filed by QUALCOMM Incorporated, San Diego, CA (US)
Filed on Dec. 19, 2022, as Appl. No. 18/067,798.
Prior Publication US 2024/0200969 A1, Jun. 20, 2024
Int. Cl. G01C 21/00 (2006.01); G05B 13/02 (2006.01); G08B 1/00 (2006.01); G08G 1/01 (2006.01)
CPC G01C 21/3807 (2020.08) [G01C 21/3841 (2020.08); G05B 13/027 (2013.01); G08G 1/0104 (2013.01)] 30 Claims
OG exemplary drawing
 
1. A device, comprising:
one or more memories; and
one or more processors, coupled to the one or more memories, configured to:
receive a set of frames of point data corresponding to sensor data associated with a vehicle,
wherein the sensor data indicates one or more sensor detections;
aggregate one or more frames of the set of frames associated with a first pose into an aggregated frame,
wherein the aggregated frame is associated with a set of cells;
obtain an indication of a respective occupancy label for each cell from the set of cells,
wherein the respective occupancy label includes a first occupancy label indicating a known occupancy status or a second occupancy label indicating an unknown occupancy status, and
wherein a subset of cells from the set of cells are associated with the first occupancy label;
convert the aggregated frame from the first pose to a second pose to generate an another aggregated frame;
train, using data associated with the another aggregated frame, a machine learning model to generate an occupancy grid,
wherein training the machine learning model is associated with a loss function that calculates a loss for respective cells from the subset of cells, and
wherein the machine learning model is trained to predict a probability of an occupancy status for respective cells from the set of cells; and
provide, to another device, the machine learning model.