A new IMM interacting approach for unequal dimension states for multitarget tracking in cluttered environments

Abstract

Since it is unknown in advance whether the target is maneuvering in practical target tracking environments, multiple model tracking techniques are introduced by applying various target dynamic models. Among the multiple model tracking techniques, interacting multiple model (IMM) method has shown excellent performance with low complexity due to the interaction process of each state of mode. When various dynamic models are designed, the dimension of each state may be unequal, which may cause biased estimate. To deal with this problem, mode interacting approaches to two dynamic models of unequal dimensions have been studied in other literatures. Here, a new interacting approach for the case of three dynamic models with unequal dimensions is proposed to reduce the bias in the extra state estimates of the higher dimensional modes, and it is shown that the tracking performance is better than the existing IMM algorithm with the conventional interaction step through Monte Carlo simulation for multi-target tracking in cluttered environments.