LiDAR Vehicle Positioning via 3D Voxel Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional positioning methods for autonomous driving, such as GPS and IMU, lack the necessary accuracy and reliability due to measurement errors and environmental limitations, making it difficult to precisely position vehicles between lanes or near obstacles.
Innovation Solution
A system and method using LiDAR data and high-definition maps that create 3-D representations of point clouds and maps through voxel matching, allowing for improved vehicle positioning by comparing these representations to determine pose information and optimize positioning accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If point-to-point matching between real-time LiDAR point clouds and high definition map is used, then positioning accuracy is improved, but computational cost increases significantly
Solution Approach 1:
The patent segments the high definition map point cloud into multiple overlapping local map point clouds, each covering a specific region. This segmentation allows the algorithm to process only relevant local regions rather than the entire map, significantly reducing computational complexity while maintaining positioning accuracy through localized point-to-point matching.
Solution Approach 2:
The patent transforms the 3D point cloud data into 2D projections by selecting specific planes (e.g., ground plane, vertical planes) for matching. This dimensionality reduction simplifies the computational burden of point-to-point matching while preserving the essential geometric features needed for accurate positioning.
2Productivity
If feature extraction and matching from segmented point clouds is used, then computational load is reduced, but positioning accuracy decreases and real-time performance is compromised
Solution Approach 1:
The patent performs point-to-point matching on selected critical planes (ground plane, vertical planes) rather than processing the entire 3D point cloud. This partial action approach focuses computational resources on the most informative subsets of data, achieving real-time performance while maintaining sufficient positioning accuracy for autonomous driving applications.
3Device complexity
If GPS and IMU positioning methods are used, then system complexity is reduced, but positioning accuracy is insufficient for autonomous driving requirements
Solution Approach 1:
The patent combines LiDAR point cloud data with high definition map data to create a hybrid positioning system. By merging real-time sensor measurements with pre-built map information, the system achieves centimeter-level positioning accuracy required for autonomous driving while maintaining reasonable system complexity through efficient point cloud processing algorithms.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances positioning accuracy and reduces computational costs, enabling real-time vehicle positioning with improved robustness against measurement noises and environmental factors.
Implementation Method 1
positioning systems and methods using Light Detection and Ranging (LiDAR)
Data Source
AI summary
Embodiments of the disclosure provide systems and methods for positioning a vehicle. The system includes a communication interface configured to receive a point cloud with respect to a scene captured by a sensor equipped on the vehicle. The system further includes a storage configured to store the point cloud and a high definition map. The system also includes a processor. The processor is configured to create a first 3-D representation of the point cloud and create a second 3-D representation of the high definition map with respect to the scene. The processor is further configured to determine pose information of the vehicle by comparing the first 3-D representation and the second 3-D representation. The processor determines a position of the vehicle based on the pose information.