Decision tree filters - a nonlinear approach to filtering for fire control applications

Abstract

A new type filter, called a Decision Tree Filter, is developed that promises substantial performance benefits in a wide variety of applications. For gun fire control against a spectrum of maneuvering and non-maneuvering targets, a relatively simple Decision Tree Filter showed a performance improvement of 2 to 1 in hit probability and a 4 to 1 reduction in computation when compared to a Kalman filter.