Constructing map data using laser scanned images

• US 10,565,682 B2
• Filed: 11/07/2016
• Issued: 02/18/2020
• Est. Priority Date: 11/07/2016
• Status: Active Grant

• Current Assignee: Ford Global Technologies, LLC (Ford Motor Company)
• Original Assignee: Ford Global Technologies, LLC (Ford Motor Company)
• Inventors: Castorena Martinez, Juan
• Primary Examiner(s): Baghdasaryan, Hovhannes
• Application Number:
  US15/345,448
• Publication Number:
  US 20180130176A1
• Time in Patent Office: 1,198 Days
• CPC Class Codes: G01S 17/42 Simultaneous measurement of distance and other co-ordinates indirect measurement G01S17/46
• CPC Class Codes: G01S 17/89 for mapping or imaging
• CPC Class Codes: G01S 7/003 Transmission of data between radar, sonar or lidar systems and remote stations
• CPC Class Codes: G01S 7/4808 Evaluating distance, position or velocity data
• CPC Class Codes: G06T 2207/20048 Transform domain processing
• CPC Class Codes: G06T 3/4038 Image mosaicing, e.g. composing plane images from plane sub-images
• CPC Class Codes: G06T 5/10 using non-spatial domain filtering
• CPC Class Codes: G06T 5/70 Denoising; Smoothing

Abstract

A computer that includes a processor and memory that stores processor-executable instructions. The processor can be programmed to: generate a plurality of transform patch images (PTPIs) by applying a mathematical transform operator to a plurality of map-perspective patch images (PMPPIs) received from at least one vehicle laser-scanner; combine at least some of the PTPIs into a single patch image; and construct a map patch image using the single patch image.

Claims

1. A computer, comprising a processor and memory that stores processor-executable instructions, wherein the processor is programmed to:

receive a plurality of map-perspective patch images of a physical region near a vehicle and imaged by at least one laser scanner on the vehicle, wherein each of the plurality of map-perspective patch images represent a measure of ground reflectivity;

generate a plurality of transform patch images by applying a mathematical transform operator to at least some of the plurality of map-perspective patch images;

determine, from among the plurality of transform patch images, at least two of the plurality of transform patch images having the largest gradients;

combine the at least two of the plurality of transform patch images into a single patch image; and

construct a map patch image using the single patch image and at least one of the plurality of map-perspective patch images.

2. The computer of claim 1, wherein the at least one laser scanner is a light detection and ranging (LiDAR) device.

3. The computer of claim 1, wherein the transform operator includes one of a discrete differentiation gradient operator, a discrete Fourier operator, a discrete wavelet operator, a Radial basis operator, a Noiselet operator, a Curvelet operator, or a Laplacian operator.

4. The computer of claim 1, wherein the processor is programmed to, in combining the at least some of the plurality of transform patch images, determine a subset of the plurality of transform patch images, wherein only the subset is combined into the single patch image.

5. The computer of claim 1, wherein the processor is programmed to, in combining the at least some of the plurality of transform patch images, execute an algorithm that includes:

6. The computer of claim 5, wherein the processor is programmed to, in combining the at least some of the plurality of transform patch images, prior to executing the algorithm, select a subset of transform patch images according to a formula, wherein only the subset is combined into the single patch image, wherein the formula includes:

7. The computer of claim 1, wherein the processor is programmed to, in combining the at least some of the plurality of transform patch images, increase signal-to-noise in the map patch image by ignoring at least some of the plurality of transform patch images.

8. The computer of claim 1, wherein the processor is programmed to, in constructing the map patch image, execute an algorithm that includes:

9. The computer of claim 1, wherein the processor is programmed generate a digital map file by combining the constructed map patch image and at least one additionally constructed map patch image generated by repeating the receiving, generating, combining, and constructing steps using a different plurality of map-perspective patch images.

10. The computer of claim 1, wherein the processor is programmed to, prior to combining the at least some of the plurality of transform patch images, apply a de-noising function to at least some of the plurality of transform patch images.

11. The computer of claim 1, wherein the processor is programmed to, in combining the at least some of the plurality of transform patch images, apply a de-noising function to the at least some of the plurality of transform patch images during fusion.

12. The computer of claim 1, wherein the processor is further programmed to, use the map patch image to operate the vehicle in an at least partially autonomous mode.

13. The computer of claim 1, wherein the computer is a remote server receiving wired or wireless data uploads from the vehicle or the computer is located within the vehicle and coupled to the at least one scanner.

14. The computer of claim 1, wherein the plurality of map-perspective patch images is captured without a calibration sequence of the at least one scanner.

15. A method, comprising:

at a computer in a vehicle;

receiving a plurality of uncalibrated map-perspective patch images of a physical region near the vehicle and imaged by at least one scanner on the vehicle, wherein each of the plurality of uncalibrated map-perspective patch images represent a measure of ground reflectivity;

generating a plurality of transform patch images by applying a mathematical transform operator to at least some of the plurality of map-perspective patch images;

determining, from among the plurality of transform patch images, at least two of the plurality of transform patch images having the largest gradients;

combining the at least two of the plurality of transform patch images into a single fused patch image; and

constructing a map patch image using the fused patch image and at least one of the plurality of map-perspective patch images.

16. The method of claim 15, wherein the transform operator includes one of a discrete differentiation gradient operator, a discrete Fourier operator, a discrete wavelet operator, a Radial basis operator, a Noiselet operator, a Curvelet operator, or a Laplacian operator.

17. The method of claim 15, wherein the combining step further comprises executing an algorithm that includes:

18. The method of claim 15, wherein the constructing step further comprises executing an algorithm that includes:

19. A computer, comprising a processor and memory that stores processor-executable instructions, wherein the processor is programmed to:

generate a plurality of transform patch images (PTPIs) by applying a mathematical transform operator to a plurality of map-perspective patch images (PMPPIs) received from at least one vehicle laser-scanner;

determine, from among the PTPIs, at least two of the PTPIs having the largest gradients;

combine the at least two of the PTPIs into a single patch image; and

construct a map patch image using the single patch image.

20. The computer of claim 19, wherein the construction includes merging both the single patch image and one of the PMPPIs.