An algorithm capable of reliably tracking a target using measurements of diverse sensors in the presence of perturbations has been presented. This algorithm is based on the analysis of the context which enable the evaluation of validity probabilities. These probabilities allow to increase the importance of measurements provided by reliable sensors and to decrease the importance of the corrupted ones in the update equations of a multisensor Kalman filter. Fuzzy logic is used for the representation of the human knowledge that describes the validity domain of the sensors. This theory is consistent with the probability theory usually used in modelling stochastic signals. Simulations results show a significant gain in performance. This gain in performance is mainly characterized by a strong robustness to natural and artificial perturbations in an operational scenario.
展开▼
