An important step in the asynchronous multi-sensor registration problem is toestimate sensor range and azimuth biases from their noisy asynchronousmeasurements. The estimation problem is generally very challenging due tohighly nonlinear transformation between the global and local coordinate systemsas well as measurement asynchrony from different sensors. In this paper, wepropose a novel nonlinear least square (LS) formulation for the problem by onlyassuming that a reference target moves with an unknown constant velocity. Wealso propose a block coordinate decent (BCD) optimization algorithm, with ajudicious initialization, for solving the problem. The proposed BCD algorithmalternately updates the range and azimuth bias estimates by solving linearleast square problems and semidefinite programs (SDPs). The proposed algorithmis guaranteed to find the global solution of the problem and the true biases inthe noiseless case. Simulation results show that the proposed algorithmsignificantly outperforms the existing approaches in terms of the root meansquare error (RMSE).
展开▼
