Autonomous Parking Start-Point Correction Using Road Mark Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous parking technologies require expensive infrastructure such as control cameras, communication systems, and Lidar, making them costly and inaccessible for widespread implementation.
Innovation Solution
A vehicle equipped with at least one camera, a display, and a storage unit for a high-definition map (HD map) of a parking lot, which generates a surround view image and corrects the starting point on the HD map based on recognized road marks, enabling autonomous parking without expensive infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If expensive infrastructure such as control cameras, communication systems, and Lidar are deployed in the parking lot, then autonomous parking reliability is improved, but device complexity and implementation cost increase
Solution Approach 1:
The vehicle performs autonomous parking using its own onboard sensors and processors without requiring external infrastructure. The vehicle's processor generates surround view images, recognizes road marks, calculates positional relationships, and controls parking operations independently, making the system self-sufficient and eliminating the need for expensive parking lot infrastructure
Solution Approach 2:
The patent replaces expensive mechanical/optical infrastructure (control cameras, Lidar) with computational image processing using the vehicle's existing camera system. By processing images through multiple magnification levels and using algorithmic road mark recognition, the system achieves reliable autonomous parking without physical infrastructure upgrades
2Ease of manufacture
If the vehicle uses onboard camera and image processing to perform autonomous parking, then implementation cost is reduced, but measurement precision of road mark position may deteriorate
Solution Approach 1:
The patent employs multi-magnification dimensionality by generating and analyzing surround view images at different magnification levels. The processor first identifies road marks in reduced magnification images for broad coverage, then uses enlarged magnification images for precise position measurement. This dimensional approach to image analysis enables accurate road mark positioning using only the vehicle's onboard camera
Solution Approach 2:
The system performs preliminary road mark detection in reduced magnification images before analyzing enlarged magnification images. By first identifying the general location and orientation of road marks through lower-resolution views, the system prepares targeted high-precision measurement, improving both efficiency and accuracy while using only onboard resources
3Manufacturing precision
If the vehicle corrects the starting point on the HD map based on recognized road mark position, then autonomous parking accuracy is improved, but processing time and computational load increase
Solution Approach 1:
The processor performs road mark recognition and starting point correction selectively rather than continuously. By activating detailed image processing and position calculation only when road marks are detected and correction is needed, the system achieves high parking accuracy while minimizing unnecessary computational overhead and processing time
Solution Approach 2:
The image processing pipeline is segmented into distinct stages: initial road mark detection in reduced magnification images, verification in enlarged magnification images, position calculation, and HD map correction. This segmentation allows the system to perform computations efficiently by processing only relevant data at each stage rather than continuously analyzing all images at full resolution
Data Source
AI summary
A vehicle for performing an autonomous parking function, the vehicle including: at least one camera configured to acquire a surrounding image of a vehicle; a display; a storage configured to store a HD map of a parking lot including an autonomous parking preparation zone; and at least one process configured to, based on the surrounding image of the vehicle being processed, generate a surround view image, based on an upper end of a road mark defining the autonomous parking preparation zone being recognized in the surround view image, track a position of the upper end of the road mark, upon the vehicle being stopped while a part of the road mark is recognized, correct a starting point on the HD map of the parking lot based on the position of the upper end of the road mark, and upon a user input for performing an autonomous parking function being received, perform the autonomous parking function based on the correct starting point.