AN IMU/MAGNETOMETER-BASED INDOOR POSITIONING SYSTEM USING KALMAN FILTERING Hendrik Hellmers, Abdelmoumen Norrdine , Jorg Blankenbach and Andreas Eichhorn Technische Universität Darmstadt, RWTH Aachen University Germany 2013 International Conference on Indoor Positioning and Indoor Navigation 임형섭
Outline Motivation Design Details Simulation Result
Motivation Improve localization accuracy IMU have drifting and non-linearity error -> combine Problem -Multipath fading -Temperature and humidity effect -Attenuation of wall and floor -People or moving objects effect Solution: Magnetometer -Penetrate building materials without attenuation, fading, multipath or signal delay
Design details MILPS(Magnetic Indoor Local Positioning System) IMU(Inertial Measurement Unit) EKF(Expanded Kalman Filter)
MILPS Reference stations consisting of magnetic coils with known positions Mobile station(user) Capturing the B of multiple coils Swiching coil’s current direction to eliminate low frequency noise Time division multiplexing Blurring effect
MILPS
IMU Consisting of 3-axial accelerometer, gyroscope, magnetometer and pressure sensor Sample rate up to 2.4kHz
IMU Position: State(for EKF):
EKF-wikipedia Prediction Update State measurement Covariance Innovation Innovation covar Kalman gain Updated state Updated covar
EKF-this paper Prediction Update State measurement Covariance Innovation Innovation covar Kalman gain Updated state Updated covar
EKF
Simulation Track starts at position 1 and end at point 13 2 Coil IMU sample rate:200Hz Magnetometer sample rate:1Hz&2Hz Average speed: 0.8m/s Mark with a small permanent magnet when passing existing track points(observation to coil A at 3,5,7,9,11) Combine MILPS & low-cost IMU(EKF) Error free&
Test bed
Result
Result
Result