Multi-sensor data fusion architecture based on adaptive Kalman filters and fuzzy logic performance assessment

In this work a novel multi-sensor data fusion (MSDF) architecture is presented. First, each measurement-vector coming from each sensor is fed to a fuzzy logic-based adaptive Kalman filter (FL-AKF); thus there are N sensors and N FL-AKFs working in parallel. The adaptation in each FL-AKF is, in the sense of dynamically tuning the measurement noise covariance matrix R, employing a fuzzy inference system (FIS) based on a covariance matching technique. A second FIS, called a fuzzy logic assessor (FLA), monitors and assesses the performance of each FL-AKF. The FLA assigns a degree of confidence, a number on the interval [0, 1], to each of the FL-AKF outputs. Finally, a defuzzification scheme obtains the fused state-vector estimate based on confidence values. The effectiveness and accuracy of this approach is demonstrated using a simulated example. Two defuzzification methods are explored and compared, and results show good performance of the proposed approach.