Road Parameter Estimation via Marker and Model Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for estimating road parameters from images captured by vehicles lack accuracy in recognizing lane information, leading to poor detection of road shapes and profiles, especially for the lane on which the vehicle is traveling.
Innovation Solution
A road parameter estimation apparatus that combines marker-based and model-based estimation methods, using a trained recognition model to recognize lane regions and demarcation lines, and determines whether to use marker-based or model-based results for more accurate feature parameter estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If marker-based estimation method is used to extract lane lines from markers, then lane line detection can be performed, but accuracy is insufficient under various road conditions and traffic environments
Solution Approach 1:
The patent combines marker-based estimation and model-based estimation methods into a unified system. The marker extractor extracts markers to generate lane lines, while the recognition model independently estimates lane lines based on image features. Both results are integrated through a selection mechanism that chooses the more accurate estimation based on comparison with ground truth data, thereby achieving both high accuracy and adaptability across various road conditions.
Solution Approach 2:
The system dynamically changes the estimation method based on road conditions and environmental parameters. By comparing the accuracy of marker-based and model-based estimations against ground truth data, the system selects the appropriate method for current conditions, effectively adapting to different scenarios such as clear weather, poor visibility, varying road markings, and different traffic environments.
2Adaptability or versatility
If model-based estimation method is used to estimate road parameters, then adaptability to various conditions is improved, but measurement precision of lane information is insufficient
Solution Approach 1:
The system incorporates feedback mechanisms where the recognition model's estimates are compared against ground truth data to evaluate accuracy. This feedback loop allows the system to determine when model-based estimation is sufficiently accurate and when marker-based extraction should be used instead, ensuring both adaptability and precision are optimized based on real-time performance evaluation.
3Device complexity
If only marker-based estimation is used, then device complexity is reduced, but reliability of road parameter estimation deteriorates
Solution Approach 1:
The system dynamically switches between marker-based and model-based estimation methods based on which provides more reliable results in current conditions. This dynamic selection mechanism, controlled by accuracy comparison with ground truth data, ensures high reliability without requiring both complex systems to operate simultaneously, thus maintaining reasonable device complexity while achieving superior reliability.
Data Source
AI summary
In a road parameter estimation apparatus, a marker-based estimator extracts, based on markers extracted by a marker extractor, at least one lane line that demarcates a road into plural regions in a width direction of the road, and estimates, based on the extracted at least one lane line, a value of at least one feature parameter of the road as a maker-based estimation result. The at least one feature parameter of the road represents at least one feature of the road. A model-based estimator estimates, based on the at least one model-based demarcation line, a value of the at least one feature parameter of the road parameter as a model-based estimation result. A determiner compares the at least one lane line with the model-based demarcation line to accordingly determine whether to use the marker-based estimation result or the model-based estimation result.