Lane Boundary Recognition for Sudden Wide-Angle Branch Roads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane boundary line recognition devices struggle to accurately detect branch roads when they suddenly branch at a wide angle from the roadway, particularly at junctions, leading to missed lane boundary lines and incorrect judgments.
Innovation Solution
A lane boundary line recognition device with a detection section and a branch judgment section that calculates the likelihood of lane boundary line candidates by monitoring changes in recognizable distance or curvature over a predetermined number of image frames, increasing the likelihood when the distance or curvature monotonically decreases, allowing for early and accurate identification of branch roads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the device uses parallel degree calculation based on estimated travel route to detect branch roads, then it can correctly identify gradual road width increases, but it fails to detect sudden wide-angle branch roads at junctions
Solution Approach 1:
The patent changes the detection parameter from parallel degree (based on estimated travel route) to recognizable distance monotonic decrease. This parameter change enables the system to detect both gradual and sudden wide-angle branch roads by monitoring whether the recognizable distance of lane boundary line candidates continuously decreases over a predetermined number of image frames, rather than relying on parallel degree calculations that fail for wide-angle branches.
2Measurement precision
If the device relies on gentle road width increase to meet branch road, then parallel degree calculation works, but sudden road width expansion at junctions causes detection failure
Solution Approach 1:
The patent applies preliminary action by monitoring the recognizable distance of lane boundary line candidates over a predetermined number of image frames before making a branch road determination. This preliminary monitoring of distance changes allows the system to prepare for and correctly identify sudden wide-angle branch roads at junctions, preventing detection failures that occur with traditional parallel degree methods.
Data Source
AI summary
An in-vehicle camera obtains image frames of a scene surrounding an own vehicle on a roadway. An extracting section in a lane boundary line recognition device extracts white line candidates from the image frames. The white line candidates indicate a degree of probability of white lines on an own vehicle lane on the roadway and a white line of a branch road which branches from the roadway. A branch judgment section calculates a likelihood of the white line as the white line of the branch road, and judges whether or not the white line candidate is the white line of the branch road based on the calculated likelihood. The branch judgment section decreases the calculated likelihood when a recognizable distance of the lane boundary line candidate monotonically decreases in a predetermined number of the image frames.