A Study on Serpentine Maneuvering Target Tracking Based on Improved Square-Root Unscented Kalman Filter

To address the significant tracking errors of conventional Kalman filters by using constant velocity (CV) or constant acceleration (CA) models during turns in highly maneuverable serpentine trajectories with frequent directional changes, this study proposes a high-order nonlinear filtering method based on the coordinate turn (CT) model. First, the jerk concept is introduced and dimensionality expansion is performed on a two-dimensional (2D) CT model to achieve precise three-dimensional (3D) modeling of serpentine maneuvers. Subsequently, an improved square-root unscented Kalman filter (SRUKF) algorithm is employed for target tracking. The proposed method adopts minimal skewness simplex sampling to reduce the number of sampling points while optimizing their distribution via proportional correction. Simulation results demonstrate that the improved algorithm maintains comparable tracking accuracy despite fewer sampling points and reduces computational time by twenty-seven percent compared to the original approach.