Camera-LiDAR Fusion Localization for Centimeter Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in robust and accurate localization in urban environments due to GPS signal unavailability and multi-path errors, leading to significant drift in visual/LiDAR SLAM methods and limited accuracy to decimeter-level precision.
Innovation Solution
A method that constructs a 3D submap using camera images and a global map from LiDAR data, extracts structured and unstructured features, matches and refines feature correspondences, and iteratively adjusts the submap location to achieve centimeter-level precision by aligning the submap with the global map using inertial navigation and probabilistic models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If visual/LiDAR SLAM methods are used for localization, then GPS signal unavailability is addressed, but localization accuracy is limited to decimeter-level precision with significant drift
Solution Approach 1:
The patent combines camera-based visual SLAM and LiDAR-based SLAM into a unified system. The camera captures images for visual feature extraction and 3D submap construction, while the LiDAR simultaneously captures depth information for complementary feature extraction and global map construction. This merging of multiple sensing modalities allows the system to achieve centimeter-level localization precision by fusing visual and depth features, overcoming the decimeter-level limitation of individual SLAM methods.
2Measurement precision
If feature matching is performed between submap and global map, then localization precision is improved, but computational complexity increases
Solution Approach 1:
The patent segments the localization process into distinct functional modules: image processing module for visual feature extraction, depth information processing module for LiDAR feature extraction, feature matching module for correspondence identification, and submap construction module for 3D map generation. Each module handles specific tasks independently, making the complex system more manageable and efficient. This segmentation allows parallel processing of visual and depth features while maintaining precise localization through coordinated feature matching between the submap and global map.
Data Source
AI summary
A method of localization for a non-transitory computer readable storage medium storing one or more programs is disclosed. The one or more programs comprise instructions, which when executed by a computing device, cause the computing device to perform by one or more autonomous vehicle driving modules execution of processing of images from a camera and data from a LiDAR using the following steps comprising: constructing a 3D submap and a global map; extracting features from the 3D submap and the global map; matching features extracted from the 3D submap against features extracted from the global map; refining feature correspondence; and refining location of the 3D submap.