A systematic approach to interactive multiple model (IMM) filter design is described, using evolution strategies (ESs), a global optimization technique inherited from the artificial intelligence field. The methodology defined is based on the definition of test scenarios and related performance goals, which are then assumed as design constraints. Then, ESs are used to search a set of filter parameters for the chosen IMM structure, trying to fulfill all design constraints, and if not possible, gracefully relaxing those constraints. The method has been successfully applied to the design of an IMM monoradar tracker for an air traffic control (ATC) application, comparing the performance of several IMM structures. The resulting filters are compared with previous designs for similar applications, with ad-hoc parameters, and with other classical optimization methods. The proposed method enables improved filter tuning in a comprehensive and robust way.
Design of IMM filter for radar tracking using evolution strategies
IEEE Transactions on Aerospace and Electronic Systems ; 41 , 3 ; 1109-1122
2005-07-01
3596433 byte
Article (Journal)
Electronic Resource
English
CORRESPONDENCE - Design of IMM Filter for Radar Tracking using Evolution Strategies
Online Contents | 2005
|CORRESPONDENCE - Improved Kalman Filter Design for Three-Dimensional Radar Tracking
Online Contents | 2001
|A Particle filter for mutistatic radar tracking
IEEE | 2014
|Kalman Filter Applications in Airborne Radar Tracking
IEEE | 1974
|