S-IMM: Switched IMM Algorithm for Maneuvering Target Tracking

For maneuvering target tracking problem, interacting multiple model (IMM) algorithm performs better than single model algorithm, however, it also has some shortcomings. This study is devoted to propose a high accurate IMM algorithm. An improved IMM algorithm, named Switched IMM(S-IMM), is proposed. The S-IMM algorithm consists of two independent IMM filters with different model sets, CV(constant-velocity)-CV-CA(constant-acceleration) and CV-CA-CA, in parallel. By comparing the sum of probabilities of non-maneuvering models in model set 1 with that of maneuvering models in model set 2, the S-IMM algorithm can decide the target motion state automatically, and select the optimal model set’s output as the final output. The simulation results verify the effectiveness of the proposed algorithm.