Kalman Filtering Versus Voting: Which Strategy is Best for Multi-Sensor Localization?

The Kalman filter has a long history in multi-sensor localization systems. Its assumption of normal distribution has however encountered difficulties in handling multipath effects for GNSS-based localization. In the meantime, the voter also has the possibility to be developed as a data fusion method, with the goal to maximize safety or availability of the system. In this paper, the Kalman filter and the voter are compared to discover their characteristics in multi-sensor localization systems. The simulation-based comparison specifically researches the improvement brought to the dependability of the system by the Kalman filter and the voter. The results show that both strategies can enhance certain aspects of dependability.