VEHICLE TRACKING USING EXTENDED KALMAN FILTER

Abstract

Kalman filter estimates the state of system form noisy sensor information. It is used to estimate the state of a linear system at various instances on the basis of the previous estimates and observations. In this thesis the subsequent position and velocity of a moving vehicle have been estimated using extended kalman filter. The dynamics of a moving vehicle are non linear and hence kalman filter directly cannot be used. Therefore, to get the response of the system an extended kalman filter has been employed. In extended kalman filter the system the non-linearity of the system had been removed using the Taylor’s series expansion and then applying the kalman filter. But if the non-linearity is too high then the extended kalman filter might fail even. It begins with an initial guess which is known as the a priori estimate. In the vehicle tracking problem it does not matter what the value of a priori estimate would be and hence a guess is made. Using this a priori value and employing ka...