US 11,693,090 B2 · Multi-domain neighborhood embedding and weighting of sampled data

Abstract

This document describes “Multi-domain Neighborhood Embedding and Weighting” (MNEW) for use in processing point cloud data, including sparsely populated data obtained from a lidar, a camera, a radar, or combination thereof. MNEW is a process based on a dilation architecture that captures pointwise and global features of the point cloud data involving multi-scale local semantics adopted from a hierarchical encoder-decoder structure. Neighborhood information is embedded in both static geometric and dynamic feature domains. A geometric distance, feature similarity, and local sparsity can be computed and transformed into adaptive weighting factors that are reapplied to the point cloud data. This enables an automotive system to obtain outstanding performance with sparse and dense point cloud data. Processing point cloud data via the MNEW techniques promotes greater adoption of sensor-based autonomous driving and perception-based systems.

Claims20

Dependent claims

A method comprising:

determining geometric properties of sampled data obtained from one or more sensors of a vehicle by at least determining geometric distances between points in the sampled data;
determining feature properties of the sampled data including feature similarities of the points;
modifying the sampled data to retain a same quantity of points that are each embedded with a respective weighting, the respective weighting being based on:
the geometric distances that are between that point and other points in the sampled data;
a sparsity of the sampled data at that point;
and
the feature similarities of that point and the other points in the sampled data;
and
responsive to modifying the sampled data, using the respective weightings embedded in the modified sampled data for enabling execution of a vehicle operation.

6 more elements

Dependencies are inferred from the claim text.

Multi-domain neighborhood embedding and weighting of sampled data

Life of this patent

Multi-domain neighborhood embedding and weighting of sampled data

Life of this patent

Abstract

Claims20