Lane Detection State Estimation for Non-Parallel Markings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane detection methods for camera-based driver assistance systems fail to provide robust and reliable lane tracking in complex traffic situations, such as construction sites, due to the assumption of parallel lane markings, leading to performance impairment, especially in autonomous vehicles.
Innovation Solution
The method involves estimating separate courses for at least two lane markings using a state estimator, determining offset values relative to a vehicle's reference axis, and assigning these values using an additional estimation method to stabilize the tracking algorithm, allowing for robust lane detection with reduced computational effort.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If a single state estimator is used to track lane markings under the assumption of parallelism, then computational effort is reduced, but reliability deteriorates in complex traffic situations where lane markings are not parallel
Solution Approach 1:
The patent segments the tracking problem into two parts: a single state estimator tracks the common course parameters (slope, curvature) of lane markings under the parallelism assumption, while a separate additional estimation method determines offset values for individual lane markings. This segmentation allows the system to maintain low computational effort while improving reliability in non-parallel situations.
Solution Approach 2:
The patent introduces offset values as an intermediary parameter that bridges the gap between the simplified parallel lane model and the actual non-parallel lane markings. These offset values allow individual lane markings to deviate from the common course while maintaining the overall structure of the single state estimator approach.
2Reliability
If separate tracking processes are used for multiple lane markings without the parallelism assumption, then reliability is improved, but computational effort increases significantly
Solution Approach 1:
The patent merges multiple tracking processes into a single state estimator by combining the common course parameters (slope, curvature) that are shared across parallel lane markings. This merging reduces computational effort while the additional offset estimation method ensures reliability by accounting for non-parallel situations.
3Device complexity
If the parallel lane marking assumption is made, then device complexity is reduced, but adaptability deteriorates in complex traffic situations
Solution Approach 1:
The patent makes the lane marking model dynamic by allowing offset values to vary for individual lane markings while maintaining a common course model. This dynamic adjustment enables the system to adapt to non-parallel lane situations without increasing the overall model complexity, as the core parallelism assumption remains but with flexible offset corrections.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for lane detection for a camera-based driver assistance system comprising the following steps: image regions in images that are recorded by a camera (14) are identified as detected lane markings (12a, 12b) if said image regions meet a specified detection criterion. At least two detected lane markings (12a, 12b) are subjected to a tracking process as lane markings (12a, 12b) to be tracked. By means of a recursive state estimator, separate progressions are estimated for at least two of the lane markings (12a, 12b) to be tracked. Furthermore, for each of a plurality of the detected lane markings (12a, 12b), a particular offset value is determined, which indicates a transverse offset of the detected lane marking (12, 12b) in relation to a reference axis (L). By means of an additional estimation method, the determined offset values are each associated with one of the separate progressions of the lane markings (12a, 12b) to be tracked.