Multi-sensor data fusion architecture

In this work we present multi-sensor data fusion architecture. The objective of the architecture is to obtain fused measured data that represent the measured parameter as accurate as possible. The architecture is based on the use of adaptive Kalman filter formed by using Kalman filter and fuzzy logic techniques. Measurements generated from each sensor are fed into an adaptive Kalman filter. So there are n adaptive Kalman filters for n sensors working in parallel. A Correlation coefficient, produced as correlating the predicted output to measured data, is used as qualifying quantity for each adaptive Kalman filter. Based on the value of the correlation coefficient the measurement noise covariance matrix was adjusted using fuzzy logic techniques. Measurements produced from these adaptive Kalman filters were fused to form a single output. Results of testing showed notable improvement for each Kalman filter over a traditional Kalman filter. Fusing data coming from several sensors showed better results than using individual sensors.