Indoor Localization of Persons in AAL Scenarios Using an Inertial Measurement Unit (IMU) and the Signal Strength (SS) from RFID Tags

Abstract

This paper presents an indoor localization system that is based on the fusion of two complementary technologies: 1) Inertial integration and 2) RFID-based trilateration. The Inertial subsystem uses an IMU (Inertial Measurement Unit) mounted on the foot of the person. The IMU approach generates a very accurate estimate of the user’s trajectory shape (limited by the drift in yaw). However, being a dead-reckoning method, it requires an initialization in position and orientation to provide absolute positioning. The IMU-based solution is updated at 100 Hz and is always available. On the other hand, the RFID-based localization subsystem provides the absolute position using the Received Signal Strength (RSS) from several long-range active tags installed in the building. Since the transmitted RF signals are subject to many propagation artifacts (reflections, absorption,...), we use a probabilistic RSS-to-Range model and a Kalman filter to estimate the position. The output of both IMU- and RFID-based subsystems are fused into one final position estimation by adaptively fitting the IMU and RFID trajectories. The integrated solution provides: absolute positioning information, a static accuracy of less than 2.3 m (in 75% of the cases) for persons at fixed positions, a smooth trajectory for moving persons with a dynamic positioning accuracy of 1.1 m (75%), a full 100% availability, and a real-time update rate of up to 100 Hz. This approach is valid for indoor navigation and particularly for Ambient Assisted Living (AAL) applications. We presented this system to the 2nd EvAAL competition (“Evaluating AAL Systems through Competitive Benchmarking”: http://evaal.aaloa.org/ ) and our CAR-CSIC system was awarded with the first prize. A detailed analysis of the experiments during the competition is presented at the end of this paper.