Aerial Navigation via Image Database Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current navigation methods for aerial vehicles in GNSS-denied areas rely on inertial measurements, which lead to increasing position uncertainty over time, and require heavy IMUs that reduce fuel capacity and range, while existing image-based systems require feature extraction and are unreliable in challenging environments.
Innovation Solution
A method that compares sensor images with geo-referenced database images to determine the aerial vehicle's position and detect GNSS interference, using normalized cross-correlation to assess image similarity and reduce uncertainty, allowing for reliable navigation without GNSS and feature extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If IMU with better components is used to lower position errors, then measurement precision is improved, but weight of moving object increases substantially
Solution Approach 1:
The patent replaces the mechanical inertial measurement system (IMU) with an optical/image-based navigation system. Instead of using accelerometers and gyroscopes to calculate position through dead reckoning, the system uses image capture devices to capture images of the environment and compares them with pre-stored images to determine position directly, eliminating the need for heavy mechanical sensors and their associated drift errors
Solution Approach 2:
The patent uses pre-stored images (copies of the environment) captured from known positions and orientations to compare against real-time images. By creating a database of reference images with associated position and orientation data, the system can determine current position through image matching without relying on cumulative inertial measurements
2Reliability
If IMU is used for navigation in GNSS-denied areas, then reliability is improved, but position uncertainty increases over time
Solution Approach 1:
The patent replaces the inertial measurement approach with an image-based approach that does not suffer from cumulative errors. Instead of integrating acceleration data over time which leads to drift, the system directly determines position by matching current images with reference images from a database, providing reliable navigation in GNSS-denied areas without increasing position uncertainty over time
Solution Approach 2:
The system continuously compares real-time images with pre-stored reference images to determine current position and orientation. This feedback mechanism allows the vehicle to correct its position estimates continuously rather than accumulating errors, maintaining measurement precision over extended periods in GNSS-denied environments
3Measurement precision
If feature extraction is used in image-based navigation, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent extracts only the essential information needed for navigation - the position and orientation data associated with pre-stored images - rather than performing complex real-time feature extraction and matching. By pre-processing and storing reference images with their metadata, the system eliminates the need for complex algorithms to identify and match features like corners, edges, or textures in real-time
Data Source
Figure 1a~1b
Figure 2
Figure 3
AI summary
The invention relates to a method (100) for navigation of an aerial vehicle. The method comprises providing (110) a sensor image from an aerial vehicle sensor. The method also comprises to repeatedly, until at least one predetermined criterion is reached (150), perform the step of setting input data (120), where the input data comprises information related to pitch angle, roll angle, yaw angle and three-dimensional position of the aerial vehicle, the step of providing (130) a two-dimensional image from a database based on the input data, where the database comprises three-dimensional geo-referenced information of the environment, and the step of comparing (140) the sensor image and the two dimensional image from the database. The method further comprises using the input data for which the two images correspond best to each other for determining (160) at least one of the following quantities pitch angle, roll angle, yaw angle and three-dimensional position of the aerial vehicle. The invention also relates to a system, a computer program and a computer program product.