Supervised multisensor tracking algorithm

We propose, in this paper, a method and an algorithm for combining symbolic and numerical information for data fusion in tracking application. The objective is to have a tracking algorithm supervised by contextual analysis. This analysis is able to detect the sensors which are reliable and those which are not in operational situations. Then, automatically the algorithm increase, in the fusion process, the importance of measurements of reliable sensor and decrease the importance of those provided by disturbed sensors. A simulation shows the effectiveness of the method and the gain in performance. This gain is essentially based on the fact that the system has a better robustness in disturbed operational situations.