Iterative Curve Fitting for Lane Line Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional lane line detection technologies struggle to accurately distinguish zebra crossings and traffic lines from lane lines due to similar visual and laser reflection characteristics, leading to incorrect decision-making in autonomous driving systems.
Innovation Solution
A method that iteratively fits curves using a fitting degree, combining three-dimensional point cloud data and image data to select candidate points for curve fitting, and iteratively updates the curve selection based on laser reflectivity and confidence levels to improve accuracy and exclude interfering lines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional lane line detection technology is used, then detection speed is maintained, but accuracy deteriorates due to inability to distinguish zebra crossings and traffic lines from lane lines
Solution Approach 1:
The patent segments the lane line detection process into multiple stages: initial curve fitting to obtain a reference curve, selection of candidate points based on the reference curve, iterative curve fitting with filtering conditions, and final lane line determination. This segmentation allows each stage to focus on specific aspects of the detection task, improving overall accuracy while managing complexity through modular processing steps.
Solution Approach 2:
The patent performs preliminary curve fitting to generate a reference curve before selecting candidate points for final lane line determination. This preliminary action establishes a baseline that guides subsequent point selection and filtering, ensuring that only points consistent with the expected lane line geometry are considered, thereby improving accuracy without requiring complex real-time analysis.
2Measurement precision
If iterative curve fitting with multiple filtering conditions is applied, then lane line detection accuracy improves, but computational time increases
Solution Approach 1:
The patent applies partial filtering by selectively evaluating candidate points against multiple conditions (distance to reference curve, laser reflectivity, image data consistency) rather than exhaustively analyzing all points. This partial action approach achieves sufficient accuracy by focusing computational resources on the most promising candidate points while avoiding unnecessary computations on clearly invalid candidates.
Solution Approach 2:
The patent implements feedback through iterative curve fitting where the reference curve generated in one iteration informs the selection of candidate points in the next iteration. This feedback mechanism allows the system to progressively refine the lane line detection by using previous results to guide subsequent analysis, improving accuracy while maintaining computational efficiency through directed rather than exhaustive searching.
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 the accuracy of lane line detection by ensuring the generated lane line has better curve characteristics and effectively differentiates lane lines from similar features like zebra crossings and traffic lines, improving the reliability of autonomous driving systems.
Implementation Method 1
combining three-dimensional point cloud data and image data to select candidate points for curve fitting, and iteratively updates the curve selection based on laser reflectivity and confidence levels
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of the present disclosure relate to a method, apparatus, device, and computer readable storage medium for determining a lane line on a road. The method includes: acquiring a first curve for use as a reference curve, the first curve being obtained by fitting a first set of points, the first set of points being selected from a plurality of sets of points in different areas of the road; iteratively executing following operations for at least one time: selecting a second set of points from the plurality of sets of points based on the reference curve; generating a second curve by fitting the second set of points, determining a fitting degree difference between a first fitting degree associated with the reference curve and a second fitting degree associated with the second curve, determining whether a convergence condition is met at least partially based on the fitting degree difference, and setting the second curve as the reference curve, in response to failing to meet the convergence condition; and determining the lane line of the road based on the second set of points when the convergence condition is met.