PHD Filter Weight Adjustment via Sensor Track ID Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current unmanned aircraft systems (UAS) face challenges in accurately tracking multiple intruder aircraft for self-separation and collision avoidance, as existing tracking systems struggle to efficiently manage and update track intensities in a probabilistic hypothesis density (PHD) filter, leading to potential inaccuracies and inefficiencies in detecting and tracking airborne objects.
Innovation Solution
The method involves generating and updating track intensities using track IDs provided by sensors, such as ADS-B and TCAS mode S, to adjust weights and determine pruning criteria, ensuring accurate tracking by comparing sensor-specific IDs and applying statistical distance tests to refine intensity weights and maintain accurate object tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If track intensities are updated using traditional PHD filter methods, then the tracking system can maintain multiple intruder aircraft tracks, but the computational complexity increases and tracking accuracy decreases due to inefficient weight management
Solution Approach 1:
The patent extracts the weight adjustment logic from the traditional PHD filter update process and implements it as a separate preprocessing step using sensor track IDs. By separating the weight determination (based on sensor ID matching) from the intensity update calculation, the system reduces computational complexity while improving tracking accuracy through more informed weight selection.
Solution Approach 2:
The patent changes the parameter used for intensity weighting from generic measurement quality metrics to sensor-specific track ID matching results. This parameter change allows the system to leverage the unique identification capabilities of individual sensors (ADS-B, TCAS, radar) to dynamically adjust track intensities, improving accuracy without proportionally increasing computational burden.
2Adaptability or versatility
If the system processes measurements from multiple sensors with different track ID systems, then it can integrate more data sources, but the difficulty of correlating measurements across sensors increases
Solution Approach 1:
The patent introduces sensor track IDs as an intermediary element that bridges different sensor measurement systems. Each sensor (ADS-B, TCAS mode S, radar) maintains its own track ID system, and the PHD filter uses these IDs as keys to correlate measurements across sensors without requiring direct integration of their different measurement formats and coordinate systems.
Solution Approach 2:
The patent segments the measurement correlation problem by handling each sensor's track IDs independently rather than attempting to unify all sensor data into a single correlation framework. The system processes measurements from different sensors separately, using their respective track ID systems, and then integrates the results through the PHD filter's intensity update mechanism.
3Reliability
If the PHD filter maintains high intensity weights for all tracks, then detection sensitivity is improved, but false alarms increase due to insufficient pruning
Solution Approach 1:
The patent implements dynamic intensity weight adjustment where track weights are not fixed but adapt based on sensor track ID matching results. Tracks that consistently appear across multiple sensors with matching IDs receive higher weights, while tracks with inconsistent or mismatched IDs receive lower weights, enabling the system to dynamically suppress potential false alarms while maintaining sensitivity to genuine targets.
Solution Approach 2:
The system uses feedback from sensor track ID matching results to adjust track intensities. The PHD filter receives feedback about which tracks are confirmed by multiple independent sensors and uses this feedback to increase their weights, while tracks lacking consistent sensor confirmation receive lower weights, automatically reducing false alarm rates while preserving detection sensitivity.
Data Source
AI summary
In one embodiment, a method for tracking multiple objects with a probabilistic hypothesis density filter is provided. The method includes comparing second track IDs corresponding to newly obtained measurements to one or more first track IDs corresponding to a Tk+1 predicted intensity having a predicted weight. If all of the one or more first track IDs match any of the second track IDs, the predicted weight is multiplied by a first value. If less than all of the one or more first track IDs match any of the second track IDs, the predicted weight is multiplied by a second value, wherein the second value is greater than the first value. The method then determines whether to prune the Tk+1 predicted intensity based on the predicted weight after multiplying with either the first value or the second value.