Moving Object Position Estimation Using 3D-to-2D Landmark Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the position of a moving object in autonomous driving systems face challenges due to the dimension mismatch between two-dimensional feature point information from a surrounding image and three-dimensional feature point information from a high-definition map, which complicates the calculation of similarity and accuracy of position estimation.
Innovation Solution
A processor-implemented method that projects three-dimensional feature point information from a high-definition map onto a two-dimensional probability map using perspective mapping, calculates similarity by summing and multiplying probabilities for each landmark, and estimates the position of the moving object using a particle filter or maximum likelihood optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional feature point information from HD map is directly compared with two-dimensional feature point information from surrounding image, then position estimation can be performed, but the dimension mismatch complicates similarity calculation and reduces estimation accuracy
Solution Approach 1:
The patent converts three-dimensional feature point information from the HD map into two-dimensional projected feature point information that can be directly compared with the two-dimensional feature point information from the surrounding image. This dimensionality transformation resolves the mismatch between 3D and 2D data, enabling accurate similarity calculation while maintaining computational feasibility.
2Ease of operation
If three-dimensional feature point information is projected onto two-dimensional probability map using perspective mapping, then dimension mismatch is resolved and similarity calculation becomes feasible, but additional processing steps are required
Solution Approach 1:
The patent introduces perspective mapping as an intermediary transformation process that converts 3D feature points into 2D projected feature points. This intermediary step acts as a bridge between the 3D HD map data and 2D image data, enabling similarity calculation while organizing the processing into manageable stages: obtaining 3D features, projecting to 2D, and calculating similarity.
Data Source
Figure 1A~1B
Figure 1C
Figure 2
AI summary
To estimate a position of a moving object, an electronic device generate two-dimensional (2D) feature point information of a landmark-based probability map from the surrounding image, obtain landmark-based three-dimensional (3D) feature point information from the HD map data, convert one of the 2D feature point information of the surrounding image to 3D or the 3D feature point information of the HD map data to 2D, determine a similarity between the converted feature point information and one of the feature point information of the surrounding image and the feature point information of the HD map data that is not converted, and estimate a position of the moving object based on the similarity.