Abstract:

For the systematic biases estimation of multi-sensor systems, an exact registration method is presented with data compression Kalman filter. First, the new method constructs the pseudomeasurements of systematic biases by the local measurements of each sensor, using unbiased converted measurement model. Then data compression strategy is introduced to get a synthetic measurement, and the Kalman filter is employed to get the estimates. As the converted model is unbiased, the estimations are consistent and robust even though the measurement noise and the systematic biases are large. Meanwhile, the data compression strategy can avoid the multiple iterations of the conventional method and a single filtering procedure is required only. Simulation results show that the new method can get a good performance and reduce the computation time cost efficiently.