Edge-Image 3D Pose Determination for Fast Vehicle Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current navigation systems, such as those using GPS and IMU, are not precise enough for determining the three-dimensional position and orientation of vehicles like UAVs or missiles near their destinations, and Structure-From-Motion algorithms are too slow for quick applications.
Innovation Solution
An image-based algorithm that converts camera images to edge images and uses a 3D evidence grid to compute the vehicle's position and attitude by combining rays from edge pixels, allowing for rapid determination of the vehicle's position and attitude relative to an object.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Structure-From-Motion algorithms are used to determine three-dimensional position and orientation, then measurement precision is improved, but productivity deteriorates due to slow computation speed
Solution Approach 1:
The patent extracts only the essential edge information from complete images, converting full images into edge images that contain only the critical geometric features needed for position and orientation determination. This extraction process removes redundant information while preserving the key structural data required for accurate measurement, thereby reducing computation time without sacrificing precision.
Solution Approach 2:
The patent segments the image processing task by dividing it into distinct stages: edge detection, ray computation, evidence grid population, and position determination. By breaking down the complex Structure-From-Motion algorithm into these manageable segments, the system can process only the essential edge features rather than analyzing complete images, thus improving computation speed while maintaining measurement accuracy.
2Device complexity
If GPS and IMU are used for navigation, then device complexity is reduced, but measurement precision deteriorates for three-dimensional position and orientation
Solution Approach 1:
The patent introduces an intermediary approach by using edge-based image processing as a mediator between simple sensor systems and complex Structure-From-Motion algorithms. The edge image processing acts as an intermediate step that provides precise position and orientation data without requiring the full complexity of traditional SfM or multiple sensors, thus improving precision while keeping the system relatively simple.
3Measurement precision
If complete images are processed to determine position and attitude, then measurement precision is improved, but loss of time increases due to processing large amounts of data
Solution Approach 1:
The patent extracts only the essential edge information from complete images, converting full images into edge images that contain only the critical geometric features needed for position and orientation determination. This extraction process removes redundant information while preserving the key structural data required for accurate measurement, thereby reducing computation time without sacrificing precision.
Solution Approach 2:
The patent applies partial action by processing only the essential edge features of images rather than analyzing complete images. This partial processing approach focuses computational resources on the most critical information (edges) while ignoring redundant data, thus achieving the necessary measurement precision with significantly reduced processing time and data volume.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of determining at least one of position and attitude in relation to an object is provided. The method includes capturing at least two images of the object with at least one camera. Each image is captured at a different position in relation to the object. The images are converted to edge images. The edge images of the object are converted into three-dimensional edge images of the object using positions of where the at least two images were captured. Overlap edge pixels in the at least two three-dimensional edge images are located to identify overlap points. A three dimensional edge candidate point image of the identified overlapped points in an evidence grid is built. The three dimensional candidate edge image in the evidence grid is compared with a model of the object to determine at least one of a then current position and attitude in relation to the object.