Fusion of information from biased sensor data by particle filtering

In this paper we address the problem of fusing information from biased sensor-data collected by a sensor network. Under the assumption that the biases of the sensors are nuisance parameters, we propose an algorithm that marginalizes them out from the estimation problem. The algorithm uses particle filtering to obtain the unknown states of the system and Kalman filtering for marginalization of the biases. We apply the proposed algorithm to the problem of target tracking using bearings-only measurements acquired by more than one sensor. The advantage of the considered method over standard particle filtering which does not assume the presence of biases is illustrated through computer simulations.