Neural Driving Corridor Detection Without Clear Road Markings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing algorithms for driver assistance systems struggle to reliably determine the driving corridor of a motor vehicle in situations where road markings are missing, complex, or temporarily obscured, such as in rural areas, construction sites, or busy urban intersections.
Innovation Solution
A method using a machine learning module with an artificial neural network that processes images from a camera to extract characteristic image features and determine the image points that delimit the driving corridor, allowing for reliable recognition even in challenging environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current image processing algorithms are used to determine the driving corridor, then the system is simple and easy to implement, but the reliability of driving corridor recognition deteriorates in situations where road markings are missing, complex, or temporarily obscured
Solution Approach 1:
The patent replaces traditional image processing algorithms with a machine learning module containing an artificial neural network. This substitution enables the system to reliably determine driving corridors in challenging environments (missing, complex, or obscured road markings) by learning from training data, thereby resolving the contradiction between reliability and complexity.
2Measurement precision
If traditional image processing algorithms are used, then the device complexity is low, but the measurement precision of the driving corridor boundaries deteriorates in challenging road conditions
Solution Approach 1:
The patent substitutes conventional image processing methods with a machine learning-based artificial neural network. This enables precise determination of driving corridor boundaries even when road markings are missing, complex, or temporarily covered, as the neural network can infer boundaries from contextual features rather than relying solely on visible markings.
3Reliability
If a single machine learning module performs both feature extraction and driving corridor determination, then the reliability of driving corridor determination is improved through integrated processing, but the device complexity increases compared to separate modules
Solution Approach 1:
The patent merges the feature extraction function and the driving corridor determination function into a single integrated machine learning module. This consolidation allows the artificial neural network to process images end-to-end, extracting relevant features and determining driving corridor boundaries in one unified operation, which improves reliability through consistent processing while managing complexity through integration.
Data Source
AI summary
The invention relates to a method for determining a driving corridor for a motor vehicle (1). The motor vehicle (10) comprises at least one camera (22) and a control unit (24), the camera (24) being designed to generate images of a front region (30) in front of the motor vehicle (10) and to forward these images to the control unit (24). The control unit (24) comprises a machine learning module (28) having an artificial neural network. The method comprises the following steps: an image of the front region (30) in front of the motor vehicle (10) is obtained by the at least one camera (22); characteristic image features of the image are extracted by means of the artificial neural network; by means of the same artificial neural network, image points (32) which delimit the driving corridor of the motor vehicle (10) and/or at least one driving corridor adjacent to the driving corridor are determined on the basis of the extracted characteristic image features. The invention also relates to a system (20) for determining a vehicle corridor.