RGB Point Cloud Map Generation for Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current point cloud registration algorithms for autonomous driving vehicles rely heavily on GPS signals, which can be unreliable in areas with tall buildings or dense forests, and are computationally complex, leading to inaccuracies and inefficiencies in high-definition 3D map generation.
Innovation Solution
A system that partitions large map areas into smaller partitions using topology information, applies optimization models to register point clouds, and merges them using a pose graph algorithm, while also synchronizing RGB images with point clouds to improve registration accuracy and reduce computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If GPS signal-based point cloud registration is used, then the system is simple to operate, but the measurement precision deteriorates in areas with tall buildings or dense forests
Solution Approach 1:
The patent introduces RGB images as an intermediary to bridge the gap between point clouds and GPS coordinates. By extracting features from RGB images and matching them with map features, the system achieves accurate localization without directly relying on GPS signals, thereby maintaining measurement precision in GPS-denied environments while keeping the system easy to operate
Solution Approach 2:
The patent replaces the GPS-based mechanical localization system with a vision-based feature matching system. Instead of relying on satellite signals that fail in urban canyons or dense forests, the system uses RGB camera images to extract visual features and match them with pre-stored map features, achieving accurate positioning without mechanical dependency on GPS
2Measurement precision
If traditional point cloud registration algorithms are used, then the measurement precision can be maintained, but the productivity deteriorates due to computational complexity
Solution Approach 1:
The patent segments the point cloud data into multiple partitions based on spatial information. Each partition is processed independently through feature extraction and matching, which reduces the computational complexity of processing entire large-scale point clouds while maintaining measurement precision through localized accurate matching
Solution Approach 2:
The patent performs preliminary feature extraction from RGB images and pre-stores map features before actual localization. This preliminary action prepares the data in advance, so during real-time operation, the system only needs to perform feature matching rather than full point cloud registration, significantly improving productivity while maintaining precision
3Measurement precision
If RGB images are synchronized with point clouds, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent makes the RGB camera serve multiple functions: it captures images for visual feature extraction, provides contextual information for point cloud registration, and enables localization in GPS-denied environments. This multi-functionality justifies the addition of RGB imaging capability while improving measurement precision without proportionally increasing device complexity
Data Source
AI summary
In one embodiment, a system receives a number of point clouds captured by one or more LIDAR sensors of an ADV and corresponding poses. The system receives a number of RGB images from one or more image capturing sensors of the ADV. The system synchronizes the RGB images with the point clouds to obtain RGB point clouds. The system extracts features from the RGB point clouds, the features including contextual and spatial information of the RGB point clouds. The system registers the RGB point clouds based on the extracted features and generates a point cloud map based on the registration of the RGB point clouds.