US 12,127,085 B2
Systems and methods for mitigating vehicle pose error across an aggregated feature map
Nicholas Baskar Vadivelu, Markham (CA); Mengye Ren, Toronto (CA); Xuanyuan Tu, Milton (CA); Raquel Urtasun, Toronto (CA); and Jingkang Wang, Toronto (CA)
Assigned to AURORA OPERATIONS, INC., Pittsburgh, PA (US)
Filed by Aurora Operations, Inc., Pittsburgh, PA (US)
Filed on Jan. 15, 2021, as Appl. No. 17/150,998.
Claims priority of provisional application 63/132,792, filed on Dec. 31, 2020.
Claims priority of provisional application 63/058,040, filed on Jul. 29, 2020.
Prior Publication US 2022/0032970 A1, Feb. 3, 2022
Int. Cl. H04W 4/46 (2018.01); B60W 60/00 (2020.01); G05D 1/00 (2006.01); G06F 18/2415 (2023.01); G06N 20/00 (2019.01); G06V 20/56 (2022.01)
CPC H04W 4/46 (2018.02) [B60W 60/00272 (2020.02); B60W 60/00276 (2020.02); G05D 1/0088 (2013.01); G05D 1/0274 (2013.01); G06F 18/2415 (2023.01); G06N 20/00 (2019.01); G06V 20/56 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, the method comprising:
obtaining sensor data associated with an environment of a first autonomous vehicle;
obtaining estimated location data indicative of a first estimated pose of a first autonomous vehicle;
determining a first intermediate environmental representation of at least a first portion of the environment of the first autonomous vehicle based, at least in part, on the sensor data;
obtaining a first message from a second autonomous vehicle, wherein the first message comprises a second intermediate environmental representation of at least a second portion of the environment of the first autonomous vehicle and second estimated location data indicative of a second estimated pose of the second autonomous vehicle;
wherein the second intermediate environmental representation is determined by the second autonomous vehicle based at least in part on second sensor data associated with the second portion of the environment of the first autonomous vehicle;
determining a first relative pose between the first autonomous vehicle and the second autonomous vehicle based, at least in part, on the intermediate environmental representation and the second intermediate environmental representation, the first relative pose identifying a displacement between the first autonomous vehicle and the second autonomous vehicle, wherein determining the first relative pose comprises:
concatenating a first plurality of encoded features of the first intermediate environmental representation with a second plurality of encoded features of the second intermediate environmental representation to generate an intermediate environmental representation pair; and
inputting the intermediate environmental representation pair into a machine-learned regression model to obtain the first relative pose;
generating an updated intermediate environmental representation based, at least in part, on the first intermediate environmental representation, the second intermediate environmental representation, and the first relative pose;
generating an autonomy output for the first autonomous vehicle based, at least in part, on the updated intermediate environmental representation; and
controlling the first autonomous vehicle based on the autonomy output for the first autonomous vehicle.