Angle-Based Target Tracking With Multi-Hypothesis Kalman Filters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Single platform sensor systems that measure angle-based information, such as direction cosines and Time-Difference Of Arrival (TDOA), have limited observability in determining whether a target is a stationary object on the Earth's surface, a moving object on the Earth's surface, or airborne, leading to inaccurate classification.
Innovation Solution
A system utilizing a Stationary Surface Filter (SSF), a Moving Surface Filter (MSF), and a Pseudorange Kalman Filter (PKF) to classify targets based on angle-based information, scoring and updating filters based on likelihood ratios, and determining the type of target by evaluating the presence of plausible Constituent Kalman Filters (CKFs).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If single platform angle-based sensor systems are used, then the system complexity is reduced, but the target classification accuracy deteriorates
Solution Approach 1:
The target classification problem is segmented into three distinct hypotheses (stationary surface, moving surface, airborne), each handled by a dedicated filter. The SSF handles stationary targets, MSF handles moving surface targets, and PKF handles airborne targets. This segmentation allows each filter to be optimized for its specific target type while maintaining overall system simplicity.
Solution Approach 2:
The Constituent Kalman Filters (CKFs) serve as intermediary computational structures that bridge the angle-based measurements and the final target classification. The CKFs are initialized based on the three hypotheses and their likelihoods are computed to determine the final classification, acting as an intermediary layer between raw measurements and classification output.
2Measurement precision
If multiple filter types (SSF, MSF, PKF) are initialized and maintained, then the target classification accuracy is improved, but the computational complexity increases
Solution Approach 1:
The system performs preliminary action by pre-initializing three distinct sets of Constituent Kalman Filters corresponding to the three target hypotheses before processing measurements. The SSF, MSF, and PKF are each initialized with appropriate state vectors and covariance matrices tailored to their specific target types, allowing the system to immediately evaluate all hypotheses without iterative initialization during tracking.
Solution Approach 2:
The system changes parameters dynamically by updating the state vectors, covariance matrices, and process noise models of each filter based on the specific characteristics of each target hypothesis. The SSF uses parameters appropriate for stationary targets, MSF for moving surface targets, and PKF for airborne targets, allowing each filter to adapt to its specific operational context.
3Reliability
If likelihood ratio thresholding is applied to update filters, then the reliability of target classification is improved, but the loss of information increases
Solution Approach 1:
The system applies partial action by using likelihood ratio thresholding to selectively update only those filters that meet the reliability criterion. Rather than updating all filters unconditionally, the system applies a threshold (e.g., 0.001) to determine which CKFs should be updated with new measurements, preventing information loss from updating unreliable filter estimates while maintaining sufficient information for accurate classification.
Data Source
AI summary
A method, system, and computer readable medium for performing a generalized angle-based tracker for a target determines angle-based information from a target signal, and initializes a Stationary Surface Filter (SSF), a Moving Surface Filter (MSF), and a Pseudorange Kalman Filter (PKF) with the angle-based information. Next, the SSF and MSF are scored by the PKF. If the information is associated, then the SSF, MSF, and PKF are updated. It is determined if at least one plausible Constituent Kalman Filter (CKF) exists for the SSF and MSF. If at least one CKF exists for the SSF and no CKF exists for the MSF, then the target is stationary surface; if at least one CKF exists for the MSF and no CKF exists for the SSF, then the target is moving surface; if no CKF exists for both the SSF and the MSF, then the target is moving airborne.