Line-of-Sight Rate Estimation Using Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for high-bandwidth guidance in intercepting moving, accelerating objects require expensive inertial-rate sensors and precise gimbals, which are not cost-effective for high-performance interception.
Innovation Solution
A method and apparatus that estimate the line-of-sight (LOS) rotation rate using imaging sensors fixed to a vehicle, eliminating the need for inertial angular body rate information by employing optical flow and image registration techniques, and utilizing inexpensive inertial rate sensors to derive the necessary guidance commands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expensive inertial-rate sensors and precision gimbals are used, then measurement precision of LOS rate is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces mechanical inertial-rate sensors and precision gimbals with an imaging sensor system that uses optical flow algorithms and image registration techniques to measure LOS rate. This substitution of mechanical measurement systems with optical-computational methods resolves the contradiction by achieving comparable measurement precision without the associated mechanical complexity and cost.
Solution Approach 2:
The patent introduces image processing algorithms (optical flow and image registration) as intermediaries between the imaging sensor and the LOS rate measurement. These computational intermediaries extract motion information from image sequences, enabling accurate LOS rate measurement without direct mechanical sensing, thus reducing device complexity while maintaining precision.
2Device complexity
If inexpensive inertial rate sensors are used, then device cost is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent replaces mechanical inertial-rate sensors entirely with an imaging-based optical flow system. This substitution eliminates the need for expensive precision inertial sensors while achieving comparable or superior measurement precision through computational methods, directly resolving the cost-precision tradeoff.
Solution Approach 2:
The patent uses imaging sensors to capture visual copies of the scene and extracts LOS rate information from these image copies through optical flow analysis. This copying approach replaces direct mechanical sensing with indirect visual measurement, reducing hardware cost while maintaining measurement precision through sophisticated image processing.
3Device complexity
If body-fixed imaging sensors are used, then device complexity is reduced, but ability to measure inertial LOS rate deteriorates
Solution Approach 1:
The patent introduces image registration algorithms as computational intermediaries that compensate for the body-fixed sensor's lack of inertial reference. By registering successive images to a reference frame and calculating optical flow, the system derives inertial LOS rate information from non-inertial image sequences, resolving the contradiction between sensor simplicity and measurement accuracy.
Solution Approach 2:
The patent uses feedback from image registration results to correct for vehicle motion and extract accurate inertial LOS rate. The registration process provides feedback about relative motion between frames, which is then used to compensate for body-fixed sensor limitations and recover inertial reference information, maintaining measurement precision despite sensor simplicity.
Data Source
AI summary
Apparatus/method estimate LOS rotation, to track, approach, pursue, intercept or avoid objects. Vehicle-fixed imagers approach/recede-from objects, recording image series with background. Computations, from images exclusively, estimate rotation vs. the vehicle, applying the estimate. Preferably, recording/estimating provide proportional navigation; scan mirrors extend strapdown-sensor FOR; applying includes measuring “range rate over range”, exclusively from interimage optical flow, using results to optimize proportional-navigation loop gain; estimating includes evaluating interframe optical flow, preregistering roughly as first approximation, selecting sequence anchor points, and applying a second, finer technique developing output registration that's a coordinate translation, aligning inertial surroundings. The approximation operates optical flow with efficient embedded registration/mapping, applying a homography matrix to nearby imagery. Alternatively, inexpensive low-quality inertial sensors establish preregistration, deriving a homography matrix. When contrast in the object direction is inadequate, dual sensors yield accurate virtual imaging with an object centroid superposed into background.