Tracking an accelerated target with a nonlinear constant heading model

This paper proposes a nonlinear model to track a maneuvering target in acceleration motion. The acceleration motion is traditionally modeled as a linear function with the state which consists of target position, speed and acceleration in the x, y and possibly z coordinates. The state elements in different coordinate are assumed uncoupled. However, This assumption is not generally true, as the state elements in different coordinates are correlated by the common target heading. Thus, a nonlinear constant heading model is suggested in this paper. To implement this nonlinear model, a two-stage least squares method is developed for track initiation, and an extended Kalman filter (EKF) and an unscented Kalman filter (UKF) are proposed to estimate the state in track maintenance. Performance of the nonlinear model is demonstrated through simulation data, and results show that the proposed nonlinear model outperforms the traditional linear model.