| CPC G06T 7/337 (2017.01) [G01S 13/89 (2013.01); G01S 17/89 (2013.01); G06T 3/06 (2024.01); G06T 3/4046 (2013.01); G06T 3/60 (2013.01); G06T 2207/10028 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/20221 (2013.01); G06T 2207/30241 (2013.01); G06T 2207/30252 (2013.01)] | 27 Claims |

|
1. A method, comprising:
generating, using at least one data processor, a fused feature map by at least concatenating a first feature map corresponding to a source point cloud and a second feature map corresponding to a target point cloud, the target point cloud corresponding to at least a portion of a same physical space as the source point cloud;
applying, using the at least one data processor, a machine learning model trained to determine, based at least on the fused feature map, a first relative transform aligning the target point cloud to the source point cloud, wherein the machine learning model determines the first relative transform by at least performing a weighted subsampling to extract, from the fused feature map, corresponding features from the source point cloud and the target point cloud such that the first feature map and the second feature map are downsampled into a combined feature map;
generating, using the at least one data processor, an aligned target point cloud by at least transforming the target point cloud in accordance with the first relative transform; and
determining, using the at least one data processor, a trajectory of a vehicle within the physical space based on at least the first relative transform.
|