Automated Parking Vision Mapping for Curved-Space Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle parking systems face challenges in accurately perceiving potential parking depths, especially in curved roads, and efficiently fusing multiple sensors' data to generate an understanding of the environmental scene, often requiring additional sensors beyond rear cameras.
Innovation Solution
A vehicle vision system utilizing CMOS cameras and advanced classification algorithms, combined with neural networking, processes and fuses sensor data to generate an annotated map for parking assistance, incorporating remote sensor data and vehicle inherent sensors, and adapts to dynamic conditions through static and dynamic reliability scoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensors are used to improve parking space detection accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system segments the sensor data processing by creating separate reliability scores for static conditions (sensor consistency) and dynamic conditions (environmental adaptability). This segmentation allows independent optimization of each scoring mechanism without increasing overall system complexity, as each score operates independently in the data fusion process.
Solution Approach 2:
The patent changes the parameter of sensor evaluation from simple presence/absence to a dual-score system (static and dynamic reliability scores). This parameter transformation enables more precise measurement of sensor quality without adding physical sensors, resolving the contradiction between measurement precision and device complexity.
2Loss of information
If additional sensors beyond rear cameras are deployed, then environmental scene understanding is improved, but device complexity and cost increase
Solution Approach 1:
The system enables existing sensors to serve multiple functions through the dual reliability scoring mechanism. The same sensor data is evaluated under both static and dynamic conditions, extracting maximum information from available sensors without requiring additional hardware. This self-service approach improves environmental understanding while avoiding increased device complexity.
Solution Approach 2:
The patent makes the sensor evaluation system universal by applying both static and dynamic reliability scores to any sensor in the system. This multi-functional evaluation framework works with rear cameras, side cameras, or any other sensors without requiring sensor-specific processing, thereby improving scene understanding without proportionally increasing system complexity.
3Measurement precision
If static and dynamic reliability scoring is implemented, then sensor data fusion accuracy is improved, but processing time increases
Solution Approach 1:
The system performs preliminary evaluation by calculating static reliability scores based on historical sensor consistency data before real-time processing. This preliminary action separates computationally intensive historical analysis from real-time operations, reducing the processing time required during actual parking maneuvers while maintaining high fusion accuracy.
Solution Approach 2:
The patent implements periodic updates of static reliability scores at predetermined intervals rather than continuous recalculation. This periodic action maintains accurate sensor evaluation while significantly reducing processing time compared to continuous computation, as the system only needs to recalculate scores at scheduled moments rather than constantly.
Data Source
AI summary
A vehicular automated parking system includes a plurality of cameras viewing exterior of the vehicle. While the vehicle is traveling along the road, the vehicular automated parking system operates in a parking space locating mode to determine available parking spaces at least in part via processing, at an electronic control unit having an image processor, of image data captured by at least one camera of the plurality of cameras. The vehicular automated parking system fuses the determined available parking spaces with map data to generate an annotated map for display at the video display screen for viewing by the driver of the vehicle to assist the driver in selecting a parking space from the determined available parking spaces. After selection of the parking space, the vehicular automated parking system parks the equipped vehicle in the selected parking space.