Vehicle Pose Estimation by Image-to-Map Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous ground vehicles face challenges in determining their position and orientation in unstructured environments due to the limitations and inaccuracies of GPS systems, particularly in environments lacking clear navigational markings.
Innovation Solution
The use of imaging sensors to capture images of the environment, which are then processed and compared to an annotated reference map to determine the vehicle's pose, utilizing techniques such as photogrammetry, line segment detectors, and machine learning systems to detect and annotate environmental features, allowing for the construction of binary maps for alignment and position determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS transceivers are used for determining positional information, then the system can operate with simple hardware, but the measurement precision and reliability deteriorate in certain environments
Solution Approach 1:
The patent introduces an intermediary system consisting of imaging sensors, feature detectors, and map matching algorithms that mediate between the vehicle and the environment. This intermediary system processes visual information to determine positional information, achieving high measurement precision without relying on GPS infrastructure.
Solution Approach 2:
The patent replaces the GPS radio frequency system with an optical-mechanical vision-based system. Instead of using electromagnetic signals from satellites, the system uses imaging sensors to capture visual data, detect environmental features, and compute position through image processing and map matching algorithms.
2Reliability
If GPS is used for navigation, then the system can operate in open environments, but the reliability deteriorates in unstructured environments lacking clear navigational markings
Solution Approach 1:
The patent changes the fundamental parameters of the navigation system from radio frequency signal processing to optical image processing. By detecting environmental features such as edges, corners, and textures in images, and matching them against stored maps, the system achieves reliable navigation in unstructured environments where GPS fails.
Solution Approach 2:
The patent employs preliminary action by pre-processing environmental images to create annotated reference maps before navigation. These maps contain detected environmental features that are stored and used for subsequent map matching, enabling the vehicle to reliably determine its position by comparing current images with the pre-created reference maps.
3Measurement precision
If imaging sensors and image processing are used to determine position, then measurement precision improves in unstructured environments, but device complexity and computational requirements increase
Solution Approach 1:
The patent applies segmentation by dividing the complex image processing task into distinct modules: imaging sensors capture images, feature detectors identify environmental features, reference maps store annotated images, and map matching algorithms compare current images with reference maps. This segmentation makes the complex system manageable and efficient.
Solution Approach 2:
The patent extracts only the essential information from images for navigation purposes. Instead of processing entire high-resolution images, the system detects and extracts environmental features such as edges, corners, and textures, and stores only these extracted features in reference maps, significantly reducing computational complexity while maintaining measurement precision.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method enables accurate determination of position and orientation in unstructured environments, reducing reliance on GPS and improving navigation accuracy, even in areas with insufficient or unreliable GPS signals.
Implementation Method 1
an imaging sensor of the autonomous ground vehicle captures an image of the environment
Implementation Method 2
The processor can then align the annotated image with the annotated reference map and determine a position and/or orientation of the autonomous ground vehicle based on the alignment
Data Source
AI summary
Described are systems and methods to utilize images to determine the position and/or orientation of a vehicle (e.g., an autonomous ground vehicle) operating in an unstructured environment (e.g., environments such as sidewalks which are typically absent lane markings, road markings, etc.). The described systems and methods can determine the vehicle's position and orientation based on an alignment of annotated images captured during operation of the vehicle with a known annotated reference map. The translation and rotation applied to obtain alignment of the annotated images with the known annotated reference map can provide the position and the orientation of the vehicle.