Methods and systems for using trained generative adversarial networks to impute 3D data for vehicles and transportation

1 . A computer-implemented method for using a trained machine learning model to improve vehicle orientation and navigation, comprising:

receiving, at one or more processors, a navigation data set comprising point cloud data relating to a terrain of an area, wherein the point cloud data includes one or more gaps;

processing, by the one or more processors, the point cloud data using the trained machine learning model to probabilistically fill the one or more gaps within the point cloud data to generate a processed navigation data set, wherein the machine learning model has been trained, at least in part, by fabricating a plurality of training data sets by extracting portions of one or more training point clouds to simulate holes in the one or more training point clouds; and

generating, by the one or more processors, a high resolution map of the terrain of the area based upon the processed navigation data set.

2 . The computer-implemented method of claim 1 , wherein the trained machine learning model is a generative adversarial network.

3 . The computer-implemented method of claim 2 , further comprising training the generative adversarial network by:

obtaining, by the one or more processors, a training set comprising a plurality of training point clouds as the one or more training point clouds;

fabricating, by the one or more processors, the plurality of training data sets by extracting the portions of each of the plurality of training point clouds to simulate the holes in the respective training point clouds, wherein the portions extracted from the plurality of training point clouds comprise a variety of shapes and sizes for each of the plurality of training point clouds; and

training, by the one or more processors, the generative adversarial network to probabilistically predict the portions extracted from the plurality of training point clouds based upon the plurality of training data sets.

4 . The computer-implemented method of claim 1 , wherein the point cloud data relating to the terrain of the area comprises a point cloud of surface elevation within the area.

5 . The computer-implemented method of claim 1 , wherein the point cloud data comprises a plurality of data types, wherein each data type includes a respective subset of the point cloud data.

6 . The computer-implemented method of claim 5 , wherein processing the point cloud data to generate the processed navigation data set comprises combining the respective subsets of the point cloud data into a combined point cloud of the processed navigation data set.

7 . The computer-implemented method of claim 1 , further comprising:

generating, by the one or more processors, a navigation decision for controlling an autonomous vehicle based upon the high resolution map.

8 . The computer-implemented method of claim 1 , further comprising:

processing, by the one or more processors, the high resolution map to determine one or more slopes of one or more portions of the terrain of the area.

9 . A computing system for using a trained machine learning model to improve vehicle orientation and navigation, comprising:

one or more processors, and

one or more memories having stored thereon computer-executable instructions that, when executed, cause the computing system to:

receive a navigation data set comprising point cloud data relating to a terrain of an area, wherein the point cloud data includes one or more gaps;

process the point cloud data using the trained machine learning model to probabilistically fill the one or more gaps within the point cloud data to generate a processed navigation data set, wherein the machine learning model has been trained, at least in part, by fabricating a plurality of training data sets by extracting portions of one or more training point clouds to simulate holes in the one or more training point clouds; and

generate a high resolution map of the terrain of the area based upon the processed navigation data set.

10 . The computing system of claim 9 , wherein the trained machine learning model is a generative adversarial network.

11 . The computing system of claim 10 , wherein the computer-executable instructions further cause the computing system to:

obtain a training set comprising a plurality of training point clouds as the one or more training point clouds;

fabricate the plurality of training data sets by extracting the portions of each of the plurality of training point clouds to simulate the holes in the respective training point clouds, wherein the portions extracted from the plurality of training point clouds comprise a variety of shapes and sizes for each of the plurality of training point clouds; and

train the generative adversarial network to probabilistically predict the portions extracted from the plurality of training point clouds based upon the plurality of training data sets.

12 . The computing system of claim 9 , wherein the point cloud data relating to the terrain of the area comprises a point cloud of surface elevation within the area.

13 . The computing system of claim 9 , wherein:

the point cloud data comprises a plurality of data types, wherein each data type includes a respective subset of the point cloud data; and

the computer-executable instructions that cause the computing system to process the point cloud data to generate the processed navigation data set cause the computing system to combine the respective subsets of the point cloud data into a combined point cloud of the processed navigation data set.

14 . The computing system of claim 9 , wherein the computer-executable instructions further cause the computing system to generate a navigation decision for controlling an autonomous vehicle based upon the high resolution map.

15 . A non-transitory computer-readable medium having stored thereon computer-executable instructions for using a trained machine learning model to improve vehicle orientation and navigation that, when executed by one or more processors of a computing system, cause the computing system to:

receive a navigation data set comprising point cloud data relating to a terrain of an area, wherein the point cloud data includes one or more gaps;

process the point cloud data using the trained machine learning model to probabilistically fill the one or more gaps within the point cloud data to generate a processed navigation data set, wherein the machine learning model has been trained, at least in part, by fabricating a plurality of training data sets by extracting portions of one or more training point clouds to simulate holes in the one or more training point clouds; and

generate a high resolution map of the terrain of the area based upon the processed navigation data set.

16 . The non-transitory computer-readable medium of claim 15 , wherein the trained machine learning model is a generative adversarial network.

17 . The non-transitory computer-readable medium of claim 16 , wherein the computer-executable instructions further cause the computing system to:

obtain a training set comprising a plurality of training point clouds as the one or more training point clouds;

fabricate the plurality of training data sets by extracting the portions of each of the plurality of training point clouds to simulate the holes in the respective training point clouds, wherein the portions extracted from the plurality of training point clouds comprise a variety of shapes and sizes for each of the plurality of training point clouds; and

train the generative adversarial network to probabilistically predict the portions extracted from the plurality of training point clouds based upon the plurality of training data sets.

18 . The non-transitory computer-readable medium of claim 15 , wherein the point cloud data relating to the terrain of the area comprises a point cloud of surface elevation within the area.

19 . The non-transitory computer-readable medium of claim 15 , wherein:

the point cloud data comprises a plurality of data types, wherein each data type includes a respective subset of the point cloud data; and

the computer-executable instructions that cause the computing system to process the point cloud data to generate the processed navigation data set cause the computing system to combine the respective subsets of the point cloud data into a combined point cloud of the processed navigation data set.

20 . The non-transitory computer-readable medium of claim 15 , wherein the computer-executable instructions further cause the computing system to generate a navigation decision for controlling an autonomous vehicle based upon the high resolution map.