Observable Degree Analysis for Multi-Sensor Fusion System.

Area of Science:

• Engineering
• Control Systems
• Signal Processing

Background:

• Multi-sensor fusion systems offer advantages like reduced error and improved filtering accuracy.
• System state observability is crucial for assessing Kalman filter convergence accuracy and speed.

Purpose of the Study:

• To evaluate different multi-sensor fusion systems from an observability perspective.
• To derive and compare the estimation error covariance for various fusion methods.
• To optimize filter performance by analyzing observability before the filtering process.

Main Methods:

• Derivation of estimation error covariance for three distinct fusion methods.
• Analysis of the observable degree for both the fusion center and local filters.
• Utilizing the Observability Degree Analysis and Eigenvalue-based Performance Metric (ODAEPM) to derive discriminant matrices.
• Mathematical proof to establish relationships among different fusion methods.

Main Results:

• The study provides discriminant matrices for the observable degree based on ODAEPM.
• Mathematical proof demonstrates the relationships and differences among fusion methods regarding observability.
• Simulation analysis using a multi-sensor constant velocity (CV) model confirms theoretical findings.

Conclusions:

• Observability is a key factor in optimizing multi-sensor fusion system performance.
• The derived mathematical framework and simulations validate the advantages of information fusion systems.
• This research provides a theoretical basis for selecting and designing effective multi-sensor fusion strategies.