Event Camera Localization Using Ground Gradient Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the location of a platform, such as satellite-based systems and image-based navigation, are inefficient in terms of computation and memory usage, especially in environments where GPS signals are unavailable, and inertial management units require frequent updates to correct drift.
Innovation Solution
A method using an airborne platform with a ground-facing event camera that captures intensity gradient signatures by moving its sightline through various directions, recording events, and comparing these signatures to a pre-stored database to estimate location, reducing computational requirements and memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image-based location determination is used to determine a first location in a large possible search space, then location accuracy is improved, but computational intensity and processing time increase significantly
Solution Approach 1:
The patent divides the image processing task into segments by using multiple processors working in parallel. Each processor handles a portion of the computational workload for comparing image features against the database, thereby maintaining high location accuracy while significantly reducing overall processing time through distributed computation.
Solution Approach 2:
The system performs preliminary actions by pre-processing and storing image features in a database before actual navigation is needed. During operation, the platform only needs to compare current sensor data against this pre-prepared database, dramatically reducing real-time computational requirements while preserving location determination accuracy.
2Measurement precision
If a comprehensive database of image features is stored for large regions, then location determination accuracy is improved, but memory storage requirements increase
Solution Approach 1:
The patent extracts only the essential and most discriminative image features from comprehensive databases for storage. Rather than storing complete high-resolution images or all possible features, the system identifies and stores only those key features that are sufficient for accurate location determination, thereby reducing memory requirements while maintaining determination accuracy.
Solution Approach 2:
The system applies local quality by storing higher-density feature data only in regions where precise location determination is most critical, while using lower-density representations in other areas. This allows the database to provide accurate location determination where needed while optimizing overall storage efficiency across the entire operational region.
3Measurement precision
If satellite-based location systems are used, then location accuracy is improved, but reliability deteriorates when signals are obstructed or jammed
Solution Approach 1:
The patent implements self-service by enabling the platform to determine its own location using onboard sensors and onboard processing of image data. The system uses its own camera or sensor array to capture images, processes these images to extract features, and compares them against the stored database to determine location independently, without relying on external satellite signals that can be obstructed or jammed.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
There is disclosed a method of estimating the location of an airborne platform, the airborne platform having a ground-facing camera having a field of view. The method comprises the steps of: moving the sightline of the camera through a range of directions; recording events triggered during motion of the sightline, each event being a change in intensity recorded at a pixel of the camera, and associating each event with a ground-referenced position; for each ground-referenced position within a patch, determining, from the recorded events, a resultant intensity gradient direction; determining a signature for the patch, which signature is the distribution of the resultant intensity gradient directions for the positions in the patch; and comparing the signature to a signature database to estimate the location of the airborne platform, the signature database comprising a number of predetermined signatures for patches at known locations.