3D Object Tracking via Weighted Least Squares Triangulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object tracking systems face challenges in accurately aligning multiple sensors' line-of-sight, leading to tracking errors due to pointing errors, especially in 3D tracking, where velocity errors are introduced, necessitating a more efficient and accurate method to determine if sensors are detecting the same object.
Innovation Solution
A robust approach using weighted least squares triangulation and standardized residuals to estimate the position and confidence of object detection, accounting for pointing errors and refraction, and correlating observations from different sensors to reduce velocity and position errors in 3D track estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple sensors are used for tracking, then tracking coverage and reliability are improved, but pointing errors cause sensors to misalign in line-of-sight, introducing velocity and position errors
Solution Approach 1:
The patent introduces an intermediary correlation algorithm that acts as a mediator between multiple sensors. This algorithm determines whether sensors are detecting the same object by analyzing observation data and calculating correlation confidence, thereby enabling coordinated tracking while compensating for pointing errors through statistical methods
Solution Approach 2:
The patent changes the parameter representation from direct sensor measurements to correlation-based confidence metrics. By transforming raw observation data into correlation confidence values and using statistical parameters to assess sensor agreement, the system resolves pointing errors through parameter transformation rather than direct measurement
2Measurement precision
If sensor pointing errors are accounted for, then correlation confidence between sensors is improved, but complex calculations are required to determine miss distance distribution
Solution Approach 1:
The patent performs preliminary characterization of sensor pointing errors before correlation assessment. By pre-determining error distributions and using these to calculate expected miss distances, the system simplifies real-time correlation calculations while maintaining high confidence accuracy
Solution Approach 2:
The patent creates a statistical model (copy) of the pointing error distribution that mirrors real sensor behavior. This model allows the system to simulate and predict miss distances without requiring complex real-time calculations, replacing intricate geometric computations with pre-computed statistical relationships
3Measurement precision
If weighted least squares triangulation is used, then position estimation accuracy is improved, but additional computational steps are required compared to simple intersection methods
Solution Approach 1:
The patent implements feedback through iterative weighted least squares triangulation. The algorithm uses standardized residuals as feedback to assess whether sensors are detecting the same object, and adjusts position estimates iteratively to minimize measurement errors, achieving high accuracy through feedback-driven refinement
Solution Approach 2:
The patent replaces simple geometric intersection methods with statistical optimization approaches. By substituting direct geometric calculations with weighted least squares optimization and statistical hypothesis testing, the system achieves superior accuracy through mathematical optimization rather than simple geometric construction
Data Source
AI summary
A method and system for tracking at least one object using a plurality of pointing sensors and a tracking system are disclosed herein. In a general embodiment, the tracking system is configured to receive a series of observation data relative to the at least one object over a time base for each of the plurality of pointing sensors. The observation data may include sensor position data, pointing vector data and observation error data. The tracking system may further determine a triangulation point using a magnitude of a shortest line connecting a line of sight value from each of the series of observation data from each of the plurality of sensors to the at least one object, and perform correlation processing on the observation data and triangulation point to determine if at least two of the plurality of sensors are tracking the same object. Observation data may also be branched, associated and pruned using new incoming observation data.