A general model-based filter initialization approach for linear and nonlinear dynamic systems

Tracking performance significantly relies on the quality of filter initialization, including accuracy and consistency, especially in a nonlinear filtering system. Most existing methods are built on assumptions that the target position varies linearly with time or the dynamic model is linear. In practical applications where the assumptions are violated, these methods may suffer from performance degradation or cannot be applied. To address this …