Object Tracking using Kalman and Particle filtering Techniques

Abstract

Object tracking has been an active field of research in the past decade. There are many challenges in tracking the object understand in which kind of system model it is moving and which type of noise it is taking . System model can be either linear or nonlinear or coordinated turn model and such noise can be either Gaussian or non-Gaussian depending on the type of filter chosen, Extended Kalman filter (EKF) is widely used for tracking moving objects like missiles, aircrafts and robots. Here analyse the instance of a single sensor or observer bearing only tracking (BOT) problem for two different models. In model 1, the target is assumed to have a constant velocity and constant course. In model 2, the target is assumed to follow a coordinated turn model with constant velocity but varying course. Extended Kalman Filter is used to track the target in both cases. For some application part, it is getting to be necessary to include components of nonlinearity and non-Gaussianity with a specific end goal to model exactly the essential dynamics of a system. The nonlinear extended Kalman filter (EKF) and the particle filter (PF) algorithms are used and compared the manoeuvring object tracking with bearing-only measurements. We also propose a proficient system for resampling particles to decrease the impact of degeneracy effect of particle propagation in the particle filter (PF) algorithm. One of the particle filter (PF) technique is sequential importance resampling (SIR). It is discussed and compared with the standard EKF through an illustrative example.