Point Cloud Map Registration for Globally Consistent HD Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current high-definition map creation methods for autonomous driving face challenges in ensuring global consistency and efficiency during the reconstruction of large areas, such as cities or countries, due to the complexity of aligning and splicing point cloud data from various sources.
Innovation Solution
A method involving the collection of point cloud data using LiDAR, GPS, and IMU, followed by association and registration of frames to create a consistent point cloud relation graph, which is then used to determine relative poses and splice the data into a high-definition map, utilizing techniques like Maximum A Posteriori estimation and distributed parallelization to improve data processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If point cloud data is collected and divided into blocks for high-definition map creation, then the map coverage area increases, but the consistency and alignment quality deteriorates
Solution Approach 1:
The patent divides the large-scale point cloud data into multiple blocks for parallel processing, which enables the system to handle large coverage areas efficiently. The segmentation is performed in a way that maintains the necessary spatial relationships between adjacent blocks, allowing for subsequent consistent alignment.
Solution Approach 2:
The patent introduces a global pose graph as an intermediary structure that connects all local block alignments. This pose graph serves as a mediator that ensures global consistency across all blocks by optimizing the relative poses between them, thereby maintaining alignment quality even as the coverage area expands.
2Manufacturing precision
If point cloud registration is performed to ensure global consistency, then the map quality improves, but the processing time increases
Solution Approach 1:
The registration process is segmented into local and global stages. Local registration is performed independently on each block to establish initial alignment, while global registration optimizes the pose graph connecting all blocks. This segmentation allows parallel processing of local registrations, significantly reducing overall processing time while maintaining global consistency.
Solution Approach 2:
The patent performs preliminary local registration on each block before conducting global optimization. This preliminary action establishes a good initial alignment that reduces the computational burden of the subsequent global optimization, thereby decreasing total processing time while ensuring final map quality.
3Productivity
If distributed parallelization is used for data processing, then the creation efficiency improves, but the system complexity increases
Solution Approach 1:
The system is segmented into independent processing units that can operate in parallel on different blocks of point cloud data. Each unit handles a specific block independently, which simplifies the implementation of parallelization while improving creation efficiency. The segmented approach naturally divides the complex task into manageable, parallelizable components.
Solution Approach 2:
The patent merges the results from multiple parallel processing units through the global pose graph optimization. This merging step consolidates the independently processed blocks into a consistent global map, managing system complexity by providing a unified framework that integrates parallel outputs without requiring complex inter-unit communication during processing.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A high-definition map creation method includes: obtaining point cloud data collected with respect to a target region, the point cloud data including K frames of point clouds and an initial pose of each frame of point cloud, K being an integer greater than 1; associating the K frames of point clouds with each other in accordance with the initial pose to obtain a first point cloud relation graph of the K frames of point clouds; performing point cloud registration on the K frames of point clouds in accordance with the first point cloud relation graph and the initial pose to obtain a target relative pose of each frame of point cloud in the K frames of point clouds; and splicing the K frames of point clouds in accordance with the target relative pose to obtain a point cloud map of the target region.