A Variable Structure Multiple Model Particle Filter for GMTI Tracking M. Sanjeev Arulampalam DSTO, Adelaide, Australia. sanjeev.arulampalam@dsto.defence.gov.au Matthew Orton Cambridge University, U.K mro20@eng.cam.ac.uk Neil Gordon Qinetiq, Malvern, U.K. njgordon@taz.qinetiq.com Branko Ristic DSTO, Adelaide, Australia branko.ristic@dsto.defence.gov.au Abstract – The problem of tracking ground targets with GMTI sensors has received some attention in the recent past. In addition to standard GMTI sensor mea- surements, one is interested in using non-standard in- formation such as road maps, and terrain-related visi- bility conditions to enhance tracker performance. The conventional approach to this problem has been to use the Variable structure IMM (VS-IMM), which uses the concept of directional process noise to model motion along particular roads. In this paper, we present a par- ticle filter based approach to this problem which we call Variable structure Multiple model Particle filter (VS- MMPF). Simulation results show that the performance of the VS-MMPF is much superior to that of VS-IMM. Keywords: GMTI Tracking, Variable Structure IMM, Particle Filter. 1 Introduction In standard tracking problems, the only inputs avail- able to the tracker are sensor measurements obtained through one or more sensors. However, in some ap- plications there may be some additional information available which could be exploited in the estimation process. For instance, one may have some knowledge of the environment in which the target is being tracked or there may be some knowledge of some constraints on the dynamic motion of the target, such as speed con- straints. An example application in a military context is Ground Moving Target Indicator (GMTI) tracking, where one may have some information of the terrain, such as road maps and visibility conditions. The ques- tion is, can this information be used by the tracker to produce better estimates of the target state? It turns out that incorporating such non-standard information in conventional Kalman filter based track- ers is not an easy task. The reason is that, in gen- eral, incorporating non-standard information leads to highly non-Gaussian posterior densities, and conven- tional trackers cannot easily handle propagation of non-Gaussian densities in a dynamic state estimation framework. However, there has been some attempt at incorporating such non-standard information within a Kalman filter based tracker. The most common of these is the Variable Structure Interacting Multiple Model (VS-IMM) algorithm [4]. The VS-IMM uses a modified version of the standard IMM, where the number, and structure, of the multi- ple models active at any particular time are allowed to vary. The various models may represent motion un- der different conditions of visibility, road constraints and target speeds. Although the VS-IMM has been shown to produce better results than methods that don’t use such non-standard information, it still has major drawbacks. In particular, the non-standard in- formation available to the tracker will lead to highly non-Gaussian posterior pdfs which are approximated by a finite mixture of Gaussians. In addition, the VS- IMM does not have a mechanism to incorporate hard constraints on position and speed. Because of these weaknesses, the use of VS-IMM has only resulted in modest improvement in accuracy over methods that do not use such non-standard information. In this project we propose a new algorithm based on Sequential Monte Carlo methods, which we term Variable Structure Multiple Model Particle Filter (VS- MMPF). The basic principle is to use particles (ran- dom samples) to represent the posterior density of the state of a target in a dynamic state estimation frame- work where non-standard information is utilised. Since particle filtering methods have no restrictions on the type of models, including the noise distributions used, 927 ISIF © 2002