Lane Boundary Detection Using Multi-Resolution Neural Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems for detecting lanes and road boundaries in autonomous vehicles are either too computationally expensive to run in real-time or suffer from inaccuracy due to shortcuts implemented to reduce computing requirements, limiting their ability to operate accurately in all road and weather conditions.
Innovation Solution
The use of a deep neural network trained with low-resolution images, region of interest images, and various ground truth masks enables accurate real-time detection of lanes and road boundaries, even in less than ideal weather and road conditions, while being computationally inexpensive.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional computer vision or machine learning techniques are used to detect lane and road boundaries, then measurement precision is improved, but productivity deteriorates due to high computational cost preventing real-time operation
Solution Approach 1:
The patent segments the image processing task by dividing the image into multiple resolutions. A low-resolution version of the image is processed first to identify candidate regions containing lane or road boundary indicators. Then, only these specific regions are processed at high resolution to detect the actual boundaries. This segmentation of processing resolution across different image regions resolves the contradiction by applying high computational effort only where necessary, achieving both real-time performance and detection accuracy.
2Productivity
If computational shortcuts are implemented to enable real-time operation, then productivity is improved, but measurement precision deteriorates due to reduced accuracy
Solution Approach 1:
The patent applies local quality by using different processing qualities for different regions of the image. Low-resolution processing is applied to the entire image for initial scanning, while high-resolution processing is applied locally only to candidate regions where lane or road boundary indicators are detected. This local quality approach ensures high detection accuracy is maintained in critical areas while achieving real-time processing through reduced computation in non-critical areas.
3Measurement precision
If high-resolution images are processed to improve detection accuracy, then measurement precision is improved, but use of energy increases due to higher computational requirements
Solution Approach 1:
The patent segments the energy-consuming processing task by processing images at low resolution first to identify candidate regions, then processing only those specific regions at high resolution. This segmentation dramatically reduces the total number of high-resolution pixels that need to be processed, thereby reducing energy consumption while maintaining detection accuracy in the critical candidate regions.
Solution Approach 2:
The patent applies high processing quality (high resolution) only locally to candidate regions rather than uniformly across the entire image. This local quality approach ensures that energy is consumed at high levels only where detection accuracy is critical, while the majority of the image is processed at lower energy cost, resolving the contradiction between accuracy and energy consumption.
Data Source
AI summary
In various examples, sensor data representative of an image of a field of view of a vehicle sensor may be received and the sensor data may be applied to a machine learning model. The machine learning model may compute a segmentation mask representative of portions of the image corresponding to lane markings of the driving surface of the vehicle. Analysis of the segmentation mask may be performed to determine lane marking types, and lane boundaries may be generated by performing curve fitting on the lane markings corresponding to each of the lane marking types. The data representative of the lane boundaries may then be sent to a component of the vehicle for use in navigating the vehicle through the driving surface.