Parking Assist Recognition Using Ground Feature Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle parking assist systems struggle to accurately park vehicles in parking lots without predefined area lines due to variations in vehicle inclination, lighting conditions, and the presence of movable objects, which can lead to misidentification of feature points, preventing autonomous parking.
Innovation Solution
A vehicle parking assist apparatus equipped with multiple cameras and an electronic control unit that registers and compares ground feature points from camera images to ensure accurate parking lot recognition, even with changing environmental conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If feature points of standing objects are registered for parking lot recognition, then the system can identify parking lots, but vehicle inclination variations and lighting changes cause misidentification of feature points
Solution Approach 1:
The system dynamically adjusts the feature point extraction process by implementing multiple extraction methods (corner detection, edge detection, region-of-interest-based extraction) and selectively applying them based on current imaging conditions. This allows the system to adapt to varying vehicle inclinations and lighting conditions, maintaining reliable parking lot recognition despite environmental changes.
Solution Approach 2:
The system changes parameters of the feature point extraction process by using different extraction algorithms and adjusting extraction regions based on detected conditions. When vehicle inclination or lighting conditions change, the system modifies its feature point extraction parameters to maintain accurate identification, thereby resolving the contradiction between reliable recognition and measurement precision.
2Device complexity
If the system uses fixed feature points for parking lot identification, then the process is simple, but it fails when environmental conditions change between registration and parking
Solution Approach 1:
The system transitions from static feature point registration to dynamic feature point extraction by implementing multiple extraction methods that can be selectively applied. This dynamic approach maintains system reliability under varying environmental conditions while keeping the overall process manageable through automated selection of appropriate extraction methods.
Solution Approach 2:
The system performs preliminary registration of parking lot information including multiple types of feature points (corner points, edge points, region-based points) during the registration phase. This preliminary action ensures that when parking occurs under different conditions, the system has pre-prepared multiple feature point representations to choose from, maintaining reliability without excessive complexity.
3Reliability
If multiple feature point extraction methods are implemented to handle environmental variations, then recognition reliability improves, but system complexity increases
Solution Approach 1:
The system implements dynamic selection among multiple feature point extraction methods based on current imaging conditions. Rather than continuously applying all methods, the system dynamically chooses the most appropriate extraction method for the current situation, thereby maintaining high reliability while managing system complexity through conditional execution.
Solution Approach 2:
The system applies different feature point extraction methods to different regions or under different conditions. For example, corner detection may be applied to specific regions while edge detection is applied to others, or extraction methods are selected based on local lighting conditions. This local quality approach maintains reliability where needed while reducing overall system complexity.
Data Source
AI summary
A vehicle parking assist apparatus acquires a camera image when the vehicle stops by a non-determined parking lot and acquires feature points of a ground of an entrance of the non-determined parking lot as non-compared entrance feature points from the camera image. The vehicle parking assist apparatus determines whether the non-determined parking lot is a registered parking lot by comparing information on the non-compared entrance feature points with registered entrance feature point information. The vehicle parking assist apparatus executes a parking assist control with using the currently-acquired parking lot information and the registered parking lot information when the vehicle parking assist apparatus determines that the non-determined parking lot is the registered parking lot.