A novel adaptive tracking algorithm for maneuvering targets

Maneuvering target tracking methods mainly include maneuver detection algorithm and adaptive tracking algorithm. Maneuver detection and adaptive tracking algorithm has its own advantages and disadvantages. The standard multi-model adaptive Kalman filter exits divergence and poor maneuver adaptation. For the above shortcoming, a modification to the standard multi-model adaptive Kalman filter is done. Multi-model parallel Kalman filter based on maneuver detection is constructed. The probability of each model being true is computed and then the weighted sum is obtained. At the same time, different state noise covariance matrix Q in each model is used. The parameter Q will be adjusted, once the target maneuver is detected in the filtering process. The simulation results show that filtering convergence speed and filtering accuracy are both improved for multi-model parallel Kalman filter based on maneuver detection. So the method is effective to improve the maneuvering targets tracking performance.