Bayes Optimal Target Estimation via Whitening Transformation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current target identification methods face challenges in handling high-dimensional sensor data with correlated measurement variates, intermittent data degradation, and the need for real-time processing, particularly in capturing both visible and hidden stochastic information while minimizing measurement failure.

Innovation Solution

A Bayes optimal estimation approach that conditions measurement variates using a stochastic system model, incorporating hidden states and measurement quality, with a dynamic mixed quadrature expression for real-time implementation, allowing for the fusion of features and optimal treatment of correlations between target type, observed features, and time-dependent target state and measurement quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If conventional target identification methods use expert knowledge to project high-dimensional state space to lower dimensional measurement space, then training sample sparsity is reduced and feature exploitation is enabled, but the calculation of posterior distribution p(T|E) becomes complicated and measurement correlations are not properly handled

Engineering Contradiction:

Improveease of target identification implementationVSAvoidaccuracy of target type estimation

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent transforms the measurement model by changing parameters from correlated raw measurements to whitened independent components through eigenvalue decomposition. This parameter transformation enables simple multiplication of likelihoods while maintaining measurement accuracy, resolving the contradiction between ease of implementation and estimation accuracy.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces complex sequential conditioning operations with a simplified parallel computation structure using the whitening transformation. Instead of sequentially conditioning on correlated measurements, the system substitutes this with independent component analysis followed by parallel likelihood calculations, greatly simplifying the computational mechanics.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If pre-screening is used to handle intermittently degraded input data, then measurement failure is managed, but the solution is sub-optimal and does not provide optimal estimation

Engineering Contradiction:

Improverobustness to data degradationVSAvoidoptimality of target estimation

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent introduces dynamic adaptation by computing the whitening transformation and likelihood functions adaptively based on current measurement conditions. This dynamic approach allows the system to optimally handle degraded data in real-time, achieving both reliability and estimation optimality unlike static pre-screening methods.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent performs preliminary whitening transformation on the measurement model to pre-compute the independent component structure and covariance characteristics. This preliminary action enables optimal handling of degraded measurements during actual operation, achieving both robustness and estimation accuracy.

Inventive Principle:

Principle #10Preliminary action

3Device complexity

If hidden states are replaced by estimated values in conventional methods, then computation is simplified, but optimal estimation and marginalization of measurement failure cannot be achieved

Engineering Contradiction:

Improvecomputational complexityVSAvoidaccuracy of target identification

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent extracts and separately handles the hidden state information through the whitening transformation, which separates the correlated measurement dependencies into independent components. This extraction allows the system to maintain computational simplicity while preserving the full information content needed for optimal estimation, avoiding the need to replace hidden states with crude estimates.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS10304001B2Robust target identification

Publication Date: 2019.05.28 RAYTHEON CO

AI summary

A target estimator that properly conditions measurement variates in the case of a series of sensor measurements collected against a target, a system model that captures visible and hidden stochastic information including but not limited to target state, target identity, and sensor measurements and that marginalizes measurement failure and a dynamic mixed quadrature expression facilitating real-time implementation of the estimator are presented.