A Modular Approach to Kalman Filter Design and Analysis
Abstract
Extended Kalman filters are important in navigation, guidance, and parameter estimation problems, and typically combine information from several different systems. This report discusses using error-states to model, linearize, and partition individual component systems. The system models are later combined into filters, which maintain the partitioning of the individual systems. This partitioning reduces the complexity of the covariance propagation and update operations. An example aided inertial navigation system is studied under two trajectories; tumble-test calibration and guided munition flight. Linearized error budgets for the munition trajectory were used to determine which sensor calibration errors it made the most sense to actively estimate, and which sensor errors contributed the most to the error in the optimal case where everything was estimated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2021
- Accession Number
- AD1126850
Entities
People
- James Maley
Organizations
- United States Army Research Laboratory