Bias fusion estimation for multi-target tracking systems with multiple asynchronous sensors

In this paper, a two-layer fusion structure is adopted to estimate the time-varying sensor bias for multi-target tracking systems with multiple asynchronous sensors. We consider the general cases, where the number of sensors is arbitrary as well as their sampling rates and initial sampling instants. First, for each target, a pseudo-measurement of sensor biases is generated by fusing all measurements of this target. In order to make the pseudo-measurement decoupled from the target state, the fusion coefficient matrix is determined to be a basis for the left null space of an augmented observation matrix. Then, without ignoring the correlations between the involved noises, a bias estimation algorithm is proposed optimally based on Kalman filter by further fusing all pseudo-measurements. The global bias estimate is proved to be unrelated to the choice of the basis for the above mentioned left-null space. Moreover, a recursive form of the proposed algorithm is provided to reduce the computational complexity. Finally, the feasibility and effectiveness of the proposed fusion estimation algorithm are illustrated by a numerical simulation.