Pedestrian Dead Reckoning with Attitude Estimation using a Fuzzy Logic Tuned Adaptive Kalman Filter

This paper presents a fuzzy logic-based pedestrian dead reckoning system relying on information derived from an inertial measurement unit (IMU) and a triaxial gyroscope. Attitude estimation is performed using a fuzzy logic tuned adaptive Kalman filter on the information fusion process. Adaptive tuning of the covariance matrices corresponding to the process and measurement noise, is carried out using a fuzzy inference system on the filter innovation sequence through a covariance-matching technique. Pedestrian walk estimation is also performed through a fuzzy logic approach which characterizes frequency and length step. Preliminary results showed an accumulate error around 6.4 % in average.