Sensor Fusion Using Detection Probability Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying and tracking a target of interest using multiple sensors cannot effectively distinguish between 'Hit' and 'False Alarm' or 'Correct Rejection' and 'Miss' from a single sensor, leading to inefficiencies in determining the true position of the target.
Innovation Solution
A method and apparatus that generate a global detection probability profile by combining detection probability profiles from multiple sensors, where 'Hits' are added and 'Correct Rejections' are subtracted, allowing for the estimation of the target's position based on probabilistic calculations and time-variant three-dimensional distributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If only sensors with 'Hit' detections are used in the solution, then the method is simple to implement, but the measurement precision deteriorates because 'Miss' and 'False Alarm' cases are ignored
Solution Approach 1:
The patent changes the parameter of detection status from binary (Hit/No Hit) to four-state (Hit, Miss, False Alarm, Correct Rejection). This allows the system to incorporate all sensor outcomes into the probability calculation, improving measurement precision by utilizing complete detection information rather than filtering out uncertain cases.
Solution Approach 2:
The patent creates a probabilistic model that copies and extends the simple Hit/No Hit detection framework to include four possible outcomes. By modeling each sensor's detection capability with probability parameters (Pd, Pfa, Pm, Pcr), the system replicates the simplicity of binary detection while adding the precision of comprehensive outcome analysis.
2Measurement precision
If all four detection outcomes are considered, then the measurement precision improves, but the device complexity increases due to the need for probability calculations
Solution Approach 1:
The patent replaces complex mechanical sensor fusion systems with a probabilistic mathematical model. Instead of using complex algorithms to process all sensor data, the system uses probability theory to calculate detection outcomes, simplifying the processing mechanism while improving precision through rigorous statistical analysis.
Solution Approach 2:
The patent simplifies the complexity by changing from processing raw sensor data to processing probability parameters. By representing sensor performance through four probability parameters (Pd, Pfa, Pm, Pcr) rather than processing all possible detection scenarios, the system reduces computational complexity while maintaining high measurement precision.
3Reliability
If sensors that lack contact with the target are excluded, then the reliability is maintained, but the productivity deteriorates due to underutilization of available sensors
Solution Approach 1:
The patent makes all sensors universal by allowing them to contribute to the solution regardless of whether they made direct contact with the target. Sensors can contribute through all four detection outcomes (Hit, Miss, False Alarm, Correct Rejection), making the system more productive by utilizing every sensor's information capability rather than requiring direct target contact.
Solution Approach 2:
The patent introduces feedback through probability calculations that incorporate all sensor outcomes. By using the detection probability profile that accounts for all four outcomes, the system continuously refines its estimate of target position, allowing sensors that lack direct contact to still contribute valuable probabilistic information to the overall solution.
Data Source
AI summary
A method and apparatus are disclosed for estimating a position of a target of interest using a plurality of position detection sensors wherein for at least one of the sensors, a corresponding time-variant detection probability profile is combined with the detection status received from the sensor and further wherein each of the corresponding detection probability profiles is combined to generate a global detection probability profile, wherein the combining comprises probabilistically adding a first given probability profile in the case where a first given corresponding indication comprises a target detection and probabilistically subtracting a second given probability profile in the case where a corresponding second given indication comprises a target non-detection and wherein the position of the target is estimated using the highest probability region(s) of the generated global detection probability profile.