Abstract: In multi-sensor bearings-only passive target tracking, the state of the target has a nonlinear relation with the bearings measurements. Existing methods focus mainly on the process of linearization. However, in this process, a precision decrease is obviously unavoidable and even filter divergence will occur so as to lose the target. Therefore a new algorithm is proposed. The state of the target is approximately estimated by least squares first which is taken as pseudo measurements for the Kalman filter, and then the IMM algorithm is employed for maneuvering target tracking. Simulation results demonstrate that the algorithm for multi-sensor maneuvering target tracking is realizable and that the tracking error is far lower than that by the existing methods.

Key words: least squares, interacting multiple model, Kalman filter, passive tracking