Automatic Parking Using Virtual Spaces When Markings Are Unclear
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic parking technologies face challenges in accurately recognizing parking spaces, especially in crowded lots, angle parking, and low light conditions, leading to inefficiencies and reduced parking efficiency.
Innovation Solution
An automatic parking method and system that generates a virtual parking space when no actual space is detected, using panoramic imaging to recognize parking lines, and allows user input for adjustments, enabling the vehicle to park automatically along planned trajectories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If panoramic image recognition is used to detect parking spaces, then the system can identify parking spaces with clear markings, but it fails to recognize parking spaces when there are many empty spaces, angle parking, or low light conditions
Solution Approach 1:
The system performs preliminary action by generating virtual parking space candidates based on detected obstacle vehicles before actual parking execution. Virtual parking spaces are pre-calculated with different types (parallel, perpendicular, angle) and positions, allowing the system to prepare multiple options in advance. This resolves the contradiction by providing predetermined solutions that don't rely on recognizing actual parking space markings, thus improving both accuracy and environmental adaptability.
Solution Approach 2:
The system creates virtual copies of parking spaces based on obstacle vehicle positions and dimensions. Instead of relying on visual recognition of actual parking space markings, the system generates virtual representations of where parking spaces could be, using mathematical models of vehicle dimensions and parking geometry. This copying approach enables the system to 'see' parking spaces that are not visually marked, resolving the contradiction between recognition accuracy and environmental adaptability.
2Measurement precision
If the system waits for clear parking space detection before initiating automatic parking, then accuracy is maintained, but parking efficiency is greatly reduced in difficult environments
Solution Approach 1:
The system performs preliminary calculation of virtual parking spaces immediately when obstacle vehicles are detected, without waiting for confirmation of actual parking space markings. This allows the parking process to initiate based on virtual candidates, significantly improving efficiency. The system maintains precision by allowing user selection and adjustment of virtual parking spaces before execution, ensuring accurate parking even in environments where visual detection fails.
Solution Approach 2:
The system provides self-service by automatically generating virtual parking space candidates and presenting them to the user for selection. When the user selects a virtual parking space, the system automatically calculates the parking trajectory and executes the parking maneuver without requiring manual intervention for trajectory planning. This self-service mechanism resolves the contradiction by enabling fast automatic parking initiation while maintaining accuracy through user confirmation.
3Adaptability or versatility
If the system generates virtual parking spaces for all possible positions, then flexibility is improved, but system complexity increases
Solution Approach 1:
The system segments the virtual parking space generation into distinct types (parallel parking, perpendicular parking, angle parking) and processes each type separately based on obstacle vehicle positions. Instead of generating all possible parking spaces uniformly, the system divides the task into structured categories with specific calculation rules for each type. This segmentation reduces computational complexity while maintaining flexibility, as each segment can be processed independently with optimized algorithms.
Solution Approach 2:
The system applies local quality by generating virtual parking spaces with different levels of detail and complexity based on local conditions. For example, parallel parking virtual spaces are generated with simpler geometry when obstacle vehicles are aligned, while angle parking generates more complex virtual spaces when vehicles are at angles. The system adjusts the granularity and complexity of virtual space generation locally based on the specific parking scenario, reducing overall system complexity while maintaining adaptability.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The present disclosure discloses an automatic parking method, device, system, and a vehicle. The automatic parking method includes: acquiring a panoramic image near a vehicle, and recognizing a parking space line in the panoramic image; determining whether a parking space exists near the vehicle according to a recognition result; if the recognition result is no, generating a virtual parking space; and controlling the vehicle to park automatically according to the virtual parking space.