High definition map updates using assets generated by an autonomous vehicle fleet

Patent metadata

Field	Value
Publication number	US-12122412-B2
Application number	US-202117173730-A
Country	US
Kind code	B2
Filing date	Feb 11, 2021
Priority date	Feb 11, 2021
Publication date	Oct 22, 2024
Grant date	Oct 22, 2024

Abstract

The disclosed technology provides solutions for updating high definition maps based on low resolution map assets. In some aspects, a process of receiving a change detection relating to a change in the real world is provided. The process can include steps for receiving autonomous vehicle drive data based on the change in the real world, generating low resolution tile data based on the autonomous vehicle drive data based on the change in the real world, generating updated semantic data based on the low resolution tile data generated, and providing the updated semantic data to an autonomous vehicle to update a proximate area of the change in the real world of a base map of the autonomous vehicle. Systems and machine-readable media are also provided.

First claim

What is claimed is:

1. A computer-implemented method comprising:
receiving, at a remote computing system, a plurality of change detection signals that include high resolution signals and low resolution signals corresponding to a real-world environment, wherein the remote computing system includes a change detection platform configured to manage and prioritize the plurality of change detection signals based on a predicted accuracy of each of the plurality of change detection signals and by modulating between the high resolution signals and the low resolution signals on a per-region basis encompassing the real-world environment;
determining, at a remote computing system, a change detection corresponding to a low priority inconsistency between a high-definition base map and the real-world environment, wherein the change detection is based on at least one change detection signal from the plurality of change detection signals that is generated by a first autonomous vehicle as the first autonomous vehicle navigates the real-world environment, and wherein the high-definition base map is based on high-resolution sensor data collected by a mapping vehicle;
receiving, at the remote computing system, two-dimensional camera image data corresponding to the real-world environment from a second autonomous vehicle, wherein the two-dimensional camera image data is collected by the second autonomous vehicle as the second autonomous vehicle navigates a proximate area of the real-world environment;
generating, by the remote computing system, low resolution, two-dimensional tile data based on the two-dimensional camera image data received from the second autonomous vehicle;
generating, by the remote computing system, updated semantic data based on the low resolution, two-dimensional tile data generated from the two-dimensional camera image data received from the second autonomous vehicle, wherein the updated semantic data is generated without utilizing corresponding high resolution tiles mapped from a redrive of the real-world environment by the mapping vehicle;
providing, by the remote computing system, the updated semantic data to the first autonomous vehicle corresponding to the real-world environment; and
directing the first autonomous vehicle to navigate the real-world environment in accordance with the updated semantic data corresponding to the real-world environment and the high-definition base map, the high-definition base map still having the low priority inconsistency between the high-definition base map and the real-world environment.

2. The computer-implemented method of claim 1 , wherein the two-dimensional camera image data includes a timestamp that is before the change detection is determined.

3. The computer-implemented method of claim 1 , wherein the two-dimensional camera image data corresponding to the real-world environment is collected by a plurality of autonomous vehicles of a fleet.

4. The computer-implemented method of claim 3 , wherein the low resolution, two dimensional tile data is stitched together using the two-dimensional camera image data collected by the plurality of autonomous vehicles to provide a view of the real-world environment, and wherein a change to one or more high resolution tiles of the high-definition base map is determined to be necessary based on the view of the real-world environment.

5. The computer-implemented method of claim 1 , further comprising designating, by the remote computing system, an avoidance area over an area proximate to the real-world environment.

6. The computer-implemented method of claim 5 , further comprising removing, by the remote computing system, the avoidance area after determining that the updated semantic data resolves the inconsistency between the high-definition base map and the real-world environment.

7. The computer-implemented method of claim 1 , wherein the updated semantic data is generated without utilizing high resolution tiles that are based on lidar data obtained from a redrive by the mapping vehicle of the real-world environment.

8. The computer-implemented method of claim 1 , further comprising:
detecting, based on the low resolution, two-dimensional tile data, one or more three-dimensional feature changes in the high-definition base map; and
in response to detecting the one or more three-dimensional feature changes, directing the mapping vehicle to perform a redrive of the real-world environment to update the high-definition base map.

9. A system comprising:
one or more processors; and
at least one computer-readable storage medium having stored therein instructions which, when executed by the one or more processors, cause the system to:
receive, by a change detection platform, a plurality of change detection signals that include high resolution signals and low resolution signals corresponding to a real-world environment, wherein the change detection platform is configured to manage and prioritize the plurality of change detection signals based on a predicted accuracy of each of the plurality of change detection signals and by modulating between the high resolution signals and the low resolution signals on a per-region basis encompassing the real-world environment;
determine a change detection corresponding to a low priority inconsistency between a high-definition base map and a real-world environment, wherein the change detection is based on at least one change detection signal from the plurality of change detection signals that is generated by a first autonomous vehicle as the first autonomous vehicle navigates the real-world environment, and wherein the high-definition base map is based on high-resolution sensor data collected by a mapping vehicle;
receive two-dimensional camera image data corresponding to the real-world environment from a second autonomous vehicle, wherein the two-dimensional camera image data is collected by the second autonomous vehicle as the second autonomous vehicle navigates a proximate area of the real-world environment;
generate low resolution, two-dimensional tile data based on the two-dimensional camera image data received from the second autonomous vehicle;
generate updated semantic data based on the low resolution, two-dimensional tile data generated from the two-dimensional camera image data received from the second autonomous vehicle, wherein the updated semantic data is generated without utilizing corresponding high resolution tiles mapped from a redrive of the real-world environment by the mapping vehicle;
provide the updated semantic data to the first autonomous vehicle corresponding to the real-world environment; and
direct the first autonomous vehicle to navigate the real-world environment in accordance with the updated semantic data corresponding to the real-world environment and the high-definition base map, the high-definition base map still having the low priority inconsistency between the high-definition base map and the real-world environment.

10. The system of claim 9 , wherein the two-dimensional camera image data includes a timestamp that is before the change detection is determined.

11. The system of claim 9 , wherein the two-dimensional camera image data corresponding to the real-world environment is collected by a plurality of autonomous vehicles of a fleet.

12. The system of claim 11 , wherein the low resolution, two-dimensional tile data is stitched together using the two-dimensional camera image data collected by the plurality of autonomous vehicles to provide a view of the real-world environment, and wherein a change to one or more high resolution tiles of the high-definition base map is determined to be necessary based on the view of the real-wor

Assignees

• Gm Cruise Holdings Llc

Inventors

• Fasola Juan
• Koduvayur Srishti
• Kushwaha Shreyans
• Lin Chia-Wei
• Lo Eugene

Classifications

• G01C21/3676

  Overview of the route on the road map · CPC title

• G08G1/20

  Monitoring the location of vehicles belonging to a group, e.g. fleet of vehicles, countable or determined number of vehicles · CPC title

• G01C21/3804 Primary

  Creation or updating of map data · CPC title

• G01C21/367

  Details, e.g. road map scale, orientation, zooming, illumination, level of detail, scrolling of road map or positioning of current position marker · CPC title

• G01S17/86

  Combinations of lidar systems with systems other than lidar, radar or sonar, e.g. with direction finders · CPC title