Road Boundary Verification Using Sourced Lateral Offset Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane support systems in autonomous and semi-autonomous vehicles face challenges in accurately and reliably tracking road boundaries, leading to safety risks and impaired user experience due to inaccurate sensor data and reliance on single reference sources.
Innovation Solution
A method that verifies road boundary representations by comparing sensor data with reference data from other vehicles, using a confidence test to ensure accuracy, and utilizing stored lateral distances to control AD or ADAS features, thereby enhancing reliability and safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a camera is mounted at a non-zero roll angle relative to the vehicle body, then the camera can be installed on vehicles with curved roofs or non-standard mounting surfaces, but the captured images contain rolled distortion that degrades ADAS and AD feature detection accuracy
Solution Approach 1:
The system performs preliminary detection of the roll angle using IMU data or image processing before the main ADAS/AD feature detection task. By identifying the roll angle in advance, the system can apply the appropriate geometric transformation to correct the distorted images, ensuring accurate feature detection despite non-zero mounting angles.
Solution Approach 2:
The system changes the geometric parameters of the image data by applying transformation matrices that compensate for the roll angle. This parameter transformation converts images captured at non-zero roll angles into an equivalent zero-roll-angle representation, maintaining detection accuracy while allowing flexible mounting positions.
2Measurement precision
If traditional calibration methods are used to correct for roll angle, then feature detection accuracy can be maintained, but the calibration process requires specialized equipment and multiple mounting positions, increasing system complexity and cost
Solution Approach 1:
The system performs self-calibration by using its own captured images and IMU data to determine the roll angle and compute the necessary geometric transformations. This eliminates the need for external calibration equipment and specialized procedures, reducing system complexity while maintaining accuracy through autonomous calibration capability.
Solution Approach 2:
The system introduces IMU data as an intermediary to facilitate the calibration process. By combining IMU-derived roll angle information with image processing, the system achieves accurate calibration without requiring complex external equipment or multiple mounting positions, simplifying the overall calibration system.
3Measurement precision
If geometric transformations are applied to correct rolled images, then feature detection accuracy is restored, but additional processing time and computational resources are required
Solution Approach 1:
The system performs preliminary determination of the roll angle using IMU data before the main image processing pipeline. By pre-computing the roll angle and preparing the transformation parameters in advance, the system minimizes the processing time required during actual ADAS/AD feature detection, reducing the time penalty associated with geometric corrections.
4Adaptability or versatility
If the camera is mounted away from the vehicle centerline to accommodate mounting constraints, then installation flexibility is improved, but lateral offset introduces additional distortion that complicates image processing
Solution Approach 1:
The system implements a universal geometric transformation framework that handles both roll angle correction and lateral offset compensation through a single integrated model. This multi-functional approach consolidates multiple correction operations into one unified processing pipeline, reducing overall system complexity while maintaining support for various mounting positions and angles.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to methods and control devices for providing a road model of a portion of a surrounding environment of a vehicle. More specifically, the present disclosure relates to utilizing stored reference data in the form of a lateral offset of a road reference (e.g. lane marker) in relation to a road boundary in order to either verify a local measurement of the lateral offset or to control a driver-assistance or autonomous driving feature based on the stored reference data. The present disclosure also relates to a method for providing verification data for road model estimations.