Innovative Hybrid Backward Input Estimation and Data Fusion for High Maneuvering Target Tracking

A hybrid unknown input estimation based on a new two-sample backward model and data fusion for high maneuvering target tracking is proposed. This new approach is based on the consideration of more than one state and input components from the current single observation. These extracted state and input components would be augmented in a single vector, and the final estimation for unknown target acceleration will be determined. Using a combination of the new backward modeling and traditional modified input estimation (MIE) technique, more information will be extracted. This new hybrid scheme which using more input information can better estimate the target maneuvering structure. Despite the traditional methods, the proposed algorithm introduces two different strategies to state the input estimation including online and delayed estimation scenarios. Also, this paper suggests several different data fusion methods through these strategies. The results are compared with a typical MIE method to evaluate the performance of the proposed hybrid scheme especially for problems in high maneuvering target tracking. The results show that the backward algorithm makes advantages such as reduction of the transient state error and more stability for the estimation by an appropriate combination of the MIE estimator.