V2I12P04

Simulation and Performance Evaluation of Unscented Kalman Filter in the Line-of-Sight Angle Tracking Problem for Different Maneuvering Targets

Trinh Thi Minh¹,  Vu Quang Luong²*, Hoang Anh Tuan², Nguyen Vu Huan²

Abstract

This paper presents a simulation study and performance evaluation of the Unscented Kalman Filter (UKF) in the problem of tracking the line-of-sight (LOS) angle for maneuvering targets with different kinematic characteristics. A third-order Singer model is used to describe the target state, including rotation angle, angular velocity, and angular acceleration. The UKF algorithm handles nonlinear and time-varying measurement noise, with the specific condition that only the rotation angle is measured under Gaussian noise. Five typical target motion types are investigated: uniform straight motion, harmonic maneuver (sinusoidal), sudden maneuver, random maneuver, and mixed maneuver. The filter quality is evaluated through plots comparing true values and estimated values for each state parameter, as well as by calculating and illustrating the mean square error (MSE) over time for each parameter. Simulation results show that UKF achieves fast convergence, stable error, and high performance under flexible and complex target maneuver conditions.

Keywords:

Unscented Kalman Filter, line-of-sight angle tracking, third-order Singer model, maneuvering target, nonlinear filtering, mean square error, tracking measurement system simulation, angular target tracking.