Recursive Nonlinear Filtering through Complete Linearization.
Abstract
Recursive filtering forms the foundation of all tracking and position estimation tasks when the target undergoes dynamic motion relative to the observer. Moreover, recursive filtering has an important role in stochastic signal detection. The typical application almost invariably involves dynamics equations which depend nonlinearly on the physical variables and sometimes even the sensor (observation) equation involves a nonlinear relation to the quantities to be estimated. In these common situations the practitioner nearly always resorts to a form of approximate linearization of the describing equations. Although highly successful in many useful application, it is commonly recognized that a variety of problems simply will not satisfactorily succumb to the usual Extended Kalman methods. For these situations present research in nonlinear filtering theory hold major portent. Indeed, the new understanding of nonlinear problems developed over the past 10 years leads to optimism for better, more refined approximate filtering techniques which may prove to be both accurate and more widely applicable.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1985
- Accession Number
- ADA155029
Entities
People
- T. F. Dyson