3D Lane Line Generation Through Scored Key-Point Connections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating three-dimensional lane lines using neural networks face inaccuracies due to mistaken detection of two-dimensional lane lines, leading to broken or disconnected lines, which reduces the accuracy of the generated three-dimensional lane lines.
Innovation Solution
A method that combines three-dimensional point cloud data and two-dimensional lane line data to determine lane line key points, utilizing attribute information and driving direction to connect these points randomly and iteratively, scoring each connection to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If two-dimensional lane lines are projected into three-dimensional space to obtain three-dimensional lane lines, then the generation process is simple, but the accuracy of the generated three-dimensional lane lines is low due to mistaken detection and broken lines
Solution Approach 1:
The patent segments the lane line generation process into multiple stages: first obtaining initial three-dimensional lane lines through projection, then identifying key points on these lines, and finally generating optimized three-dimensional lane lines by connecting key points. This segmentation allows each stage to focus on specific tasks, improving overall accuracy while maintaining processability.
Solution Approach 2:
The patent implements feedback by using the initial three-dimensional lane lines (generated from two-dimensional projections) as input for identifying key points, which then serve as basis for generating optimized three-dimensional lane lines. This iterative feedback process continuously refines the lane line accuracy, resolving the contradiction between simple generation and high precision.
2Productivity
If two-dimensional lane lines are detected and projected to generate three-dimensional lane lines, then the processing speed is fast, but the reliability of the lane line data is reduced due to mistaken detection
Solution Approach 1:
The patent performs preliminary action by first obtaining initial three-dimensional lane lines through projection of two-dimensional lane lines, then identifies key points on these preliminary lines before generating the final optimized lines. This preliminary processing maintains speed while setting up a foundation for improved reliability through subsequent key point-based refinement.
Solution Approach 2:
The patent uses copying by creating multiple candidate three-dimensional lane lines through random connection of key points (as mentioned in claim 1, step 3), then selects the most reliable one based on scoring. This copying approach maintains processing speed while improving reliability through selection from multiple versions.
3Measurement precision
If lane line key points are connected randomly for many times to generate multiple three-dimensional lane lines, then the accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent applies parameter changes by using scoring mechanisms to evaluate and select the best three-dimensional lane lines from multiple random connections. Instead of exhaustively checking all possibilities, the system changes parameters (scoring criteria) to efficiently identify high-accuracy lines, reducing computational complexity while maintaining precision.
Solution Approach 2:
The patent uses partial action by performing random connection only a limited number of times (not all possible combinations) and selecting the best result. This partial excessive action approach improves accuracy sufficiently while controlling computational complexity, avoiding the need to process every possible lane line configuration.
Data Source
AI summary
A method for generating a lane line, and includes: acquiring three-dimensional point cloud data and two-dimensional lane line data of a target lane, the two-dimensional lane line data being lane line data collected in response to determining that the vehicle is driving in the target lane, and the three-dimensional point cloud data being point cloud data of the surrounding environment of the vehicle collected in response to determining that the vehicle is driving in the target lane; determining a plurality of lane line key points according to the two-dimensional lane line data and the three-dimensional point cloud data; and obtaining a target three-dimensional lane line by connecting the plurality of lane line key points randomly for many times according to attribute information of the plurality of lane line key points and a driving direction of the vehicle.