On adaptive sampling for multisensor tracking of a maneuvering target using IMM/PDA filtering

We present a novel adaptive sampling scheme for tracking a highly maneuvering target. During the maneuvering mode of a target, estimation accuracy of the tracker is typically worse than that during the constant velocity (benign motion) mode. Hence it is desired to have higher sampling rate during the maneuvering mode and relatively lower sampling rate for the benign motion mode. In the proposed adaptive sampling policy, we predict an expected cost function for a candidate sampling interval selected from a predefined set of sampling intervals. We then choose the largest sampling interval for which an expected performance metric is satisfied. We use a random sampling approach coupled-with switching multiple kinematic models for target motion, to generate future (pseudo-)measurements which allows computation of the relevant performance metric. The algorithm is illustrated via a simulation example involving tracking of a maneuvering target in presence of clutter and two sensors (radar and sonar). Multisensor tracking is achieved by a suboptimal filtering algorithm developed by the interacting multiple model (IMM) filtering approach combined with the probabilistic data association (PDA) technique and with the proposed adaptive sampling scheme.