Autonomous Parking Spot Endpoint Detection for Precise Vehicle Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous parking systems for motor vehicles require operator intervention to locate and position the vehicle for parking, and there is a need for improved technologies that utilize preexisting infrastructure to enhance accuracy and robustness without increasing costs.
Innovation Solution
A computer-based system using a processor and sensors to classify and determine endpoints of parking spots through convolutional neural networks, allowing for autonomous alignment and actuation of vehicle systems to park the vehicle accurately, even in non-standard parking configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If the operator manually locates and positions the vehicle for parking, then the parking process can be completed, but the automation level remains low and operator time is consumed
Solution Approach 1:
The system enables the vehicle to park itself autonomously by detecting parking spots using sensors and convolutional neural networks, determining spot dimensions and endpoints automatically, and controlling vehicle actuation systems without operator intervention for spot identification and positioning
Solution Approach 2:
The patent replaces manual operator actions with an automated system comprising sensors, processors executing convolutional neural networks, and vehicle actuation systems, substituting human mechanical operations with electronic detection and control mechanisms
2Measurement precision
If traditional parking assistance systems are used, then basic parking aid is provided, but accuracy in determining parking spot dimensions is insufficient
Solution Approach 1:
The system transitions from simple 2D image capture to 3D spatial understanding by determining both x and y coordinates of parking spot endpoints, calculating actual dimensions through coordinate differences, and using average length comparisons to correct detection errors
Solution Approach 2:
The system employs feedback mechanisms by comparing detected parking spot lengths against average lengths from multiple detections, identifying and correcting erroneous endpoint classifications, and iteratively improving measurement accuracy through statistical validation
3Extent of automation
If autonomous parking systems are implemented, then parking automation is achieved, but system cost increases
Solution Approach 1:
The system achieves multiple functions using a unified approach: the same sensor array and convolutional neural network infrastructure performs both object detection and dimensional measurement, while the processing system handles both endpoint classification and coordinate calculation, reducing overall system cost through functional integration
4Reliability
If standard endpoint classification is used, then basic detection is achieved, but robustness against non-standard parking configurations is insufficient
Solution Approach 1:
The system adapts to non-standard parking configurations by dynamically adjusting detection parameters, using average length calculations to identify anomalies in real-time, and correcting endpoint coordinates based on statistical deviations from expected dimensional ranges
Data Source
AI summary
A system is provided that includes a computer including a processor and a memory. The memory includes instructions such that the processor is programmed to: receive an image depicting a parking spot, determine a length of the parking spot based on a classified endpoint of the parking spot, compare the length to an average length, and determine an endpoint of the parking spot when the length is less than the average length, wherein the determined endpoint is distal to the classified endpoint.