Lane Estimation Using Road Region Shape Without Lane Markers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane estimation techniques rely heavily on lane markers, which can be inaccurate in cases of construction, deterioration, or obstruction, leading to unreliable lane recognition.
Innovation Solution
A device and method that estimate lanes by recognizing shapes indicating road regions from image data, calculating feature values such as inclination angles, center-of-gravity coordinates, and pixel values, and determining threshold ranges to identify the lane without relying on lane markers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lane markers are used for lane estimation, then the lane can be recognized under normal conditions, but the lane cannot be accurately estimated when lane markers are damaged, faded, or obscured by construction repair marks
Solution Approach 1:
The patent segments the lane recognition task into multiple independent feature extraction components: road surface texture analysis, shadow region detection, lane marker pattern recognition, and geometric shape analysis. By dividing the recognition system into these separate segments, the system can rely on alternative features (such as shadow patterns or road texture) when lane markers are damaged or obscured, thereby maintaining reliability across diverse road conditions.
Solution Approach 2:
The patent changes the parameters used for lane estimation from solely relying on lane marker presence to incorporating multiple parameters including shadow region characteristics, road surface texture features, and geometric shape parameters. This parameter diversification allows the system to adapt to varying road conditions where traditional lane markers may be unreliable, thus improving both reliability and adaptability.
2Ease of manufacture
If traditional lane marker detection is used, then the system is simple to implement, but it fails when lane markers are absent or illegible
Solution Approach 1:
The patent creates a universal lane estimation system that performs multiple functions: it can detect lane markers when present, analyze shadow regions when markers are absent, examine road surface textures, and utilize geometric shape analysis. This multi-functional approach maintains ease of implementation through a unified processing framework while significantly improving reliability by having multiple fallback mechanisms when primary lane markers are unavailable or illegible.
3Device complexity
If the system focuses on lane marker detection, then processing is straightforward, but it cannot handle roads with construction marks or deteriorated markers
Solution Approach 1:
The patent transitions from two-dimensional lane marker pattern recognition to incorporating three-dimensional geometric analysis of shadow regions and road surface features. By adding this spatial dimension, the system can distinguish between actual lane markers and construction repair marks based on their geometric properties and shadow characteristics, thereby improving adaptability to various road conditions without excessively increasing processing complexity.
Data Source
AI summary
A technology is provided whereby a lane can be correctly estimated without relying on lane markers. A shape indicating a road region is extracted from image data obtained by capturing images in the progression direction of a vehicle 6 and, on the basis of information indicating the shape, the angle of inclination for one contour line, the center-of-gravity coordinates for a diagram indicating the shape of the road region, the angle between two consecutive contour lines across one vertex of the diagram indicating the shape, and the area of the diagram indicating the shape are calculated as feature values for the shape of the road region. In addition, the lane in which the vehicle 6 is traveling can be estimated by determining whether or not the calculated feature values are included in a threshold value range preset for each lane.