Tracking systems are based on models, in particular, the target dynamics model and the sensor measurement model. In most practical situations the two models are not known exactly and are typically parametrized by an unknown random vector \theta. The paper proposes a Bayesian algorithm based on importance sampling for the estimation of the static parameter \theta. The input are measurements collected by the tracking system, with non-cooperative targets present in the surveillance volume during the data acquisition. The algorithm relies on the particle filter implementation of the probability density hypothesis (PHD) filter to evaluate the likelihood of \theta. Thus, the calibration algorithm, as a byproduct, also provides a multi-target state estimate. An application of the proposed algorithm to translational sensor bias estimation is presented in detail as an illustration. The resulting sensor-bias estimation method is applicable to asynchronous sensors and does not require prior knowledge of measurement-to-target associations.
↧