Mixed-Domain Parking Zone Mapping for GPS-Denied Vehicle Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current parking assist systems rely on GPS and low-cost sensors, which are not robust in environments with weak or no GPS signal, such as rural areas or underground parking, and are influenced by lighting conditions and moving objects, leading to inefficient parking zone mapping and localization.
Innovation Solution
A mixed-domain neural network system that uses image-domain and bird's-eye-view (BEV) domain data from vehicle cameras for localization and mapping, integrating both domains without the need for GPS or inertial measurement units, while allowing for the use of these data if available to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If GPS and low-cost sensors are used for parking assist, then device complexity is reduced, but reliability deteriorates in environments with weak or no GPS signal
Solution Approach 1:
The patent replaces GPS-based mechanical positioning systems with a vision-based neural network system that processes image data to achieve localization and mapping, eliminating dependency on GPS signals and improving reliability in GPS-denied environments
Solution Approach 2:
The patent combines multiple data domains (image-domain data and BEV-domain data) to create a composite sensing approach that leverages the strengths of each domain while compensating for their individual weaknesses, achieving robust localization and mapping
2Ease of operation
If GPS-based systems are used, then ease of operation is improved, but measurement precision deteriorates in challenging environments
Solution Approach 1:
The patent substitutes GPS-based positioning with a neural network-based visual positioning system that processes camera images to determine vehicle location and orientation, achieving high precision in environments where GPS fails
Solution Approach 2:
The patent transforms image data into BEV-domain representations and processes multiple parameters (image-domain features, BEV features, localization data) through the neural network to achieve precise localization and mapping outputs
3Measurement precision
If image processing is enhanced for better detection, then measurement precision improves, but use of energy increases
Solution Approach 1:
The patent segments the processing into two distinct domains (image-domain and BEV-domain) with specialized neural network branches, allowing efficient processing of different feature types and reducing overall computational complexity while maintaining high detection precision
Data Source
AI summary
Methods and systems for assisting a vehicle to park using mixed-domain image data. Image-domain data is generated based on raw image data received from a plurality of cameras mounted on a vehicle. The raw image data is associated with a parking zone outside the vehicle, and the image-domain data is generated by a feature-detection machine learning model. A bird's-eye-view (BEV) image is generated based on the raw image data, wherein the BEV image is a projected image of the parking zone. BEV-domain data associated with the BEV image is generated. The BEV-domain data includes data associated with parking landmarks in the parking zone. A computing system localizes the vehicle within the parking zone based on the BEV-domain data and the image-domain data to generate localization data. The computing system performs mapping of the parking zone based on the BEV-domain data, the image-domain data, and the localization data.