Point Cloud Update via Pose-Based Area Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for updating historical point clouds in high-precision maps for autonomous driving are inefficient due to high operation amounts, leading to long time consumption.
Innovation Solution
The method determines the area point cloud matching newly added point clouds based on the pose of the point cloud acquisition device, allowing only the matching area point cloud to participate in the splicing process, reducing the overall operation amount and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all historical point clouds are used for splicing when updating, then the completeness of map information is improved, but the operation amount increases leading to low efficiency and long time consumption
Solution Approach 1:
The patent segments the historical point clouds into multiple area point clouds based on geographical regions. When updating, only the area point clouds corresponding to the region where new point clouds were acquired are selected for splicing, rather than processing all historical point clouds. This segmentation approach maintains map information completeness for updated regions while significantly reducing the operation amount and processing time.
2Productivity
If the scope of point cloud splicing is reduced to only matching area point clouds, then the operation amount is reduced improving efficiency, but the coverage of updated information may be limited
Solution Approach 1:
The patent uses the pose information (position and orientation) of the point cloud acquisition device as an intermediary to establish correspondence between new point clouds and historical area point clouds. By matching based on pose data, the system accurately identifies which area point clouds should be updated, ensuring that the reduced scope of splicing still covers all necessary regions without gaps in information coverage.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of the present disclosure provide a method and apparatus for updating a point cloud, and relate to the field of autonomous driving. A specific implementation includes: determining an area point cloud matching newly added point clouds from historical point clouds to be updated based on a set formed by a pose of a point cloud acquisition device acquiring the newly added point clouds; merging the newly added point clouds into the matched area point cloud to obtain merged historical point clouds; and performing global optimization on the merged historical point clouds based on a pose of each point cloud in the merged historical point clouds to obtain updated point clouds. The pose of the point cloud acquisition device is used for determining the area point cloud matching the newly added point clouds, and only the area point cloud matching the newly added point clouds is called, without requiring all the to historical point clouds to participate in the operation, to realize the splicing of the newly added point clouds, so that the calculation amount during the splicing of the point clouds is effectively reduced, and the efficiency of updating the point cloud is improved.