AN INDOOR POSITIONING SYSTEM BASED ON COMBINED AUDIO CHIRP/MEMS/FLOOR MAP: PERFORMANCE ANALYSIS OF KEPLER A100

Abstract

In recent years, audio positioning technology has been more favoured because it is not only highly accurate but also universally applicable to the mass users. In line with this trend, a chip dedicated to indoor positioning, the Kepler A100, has been released lately, which ranges by estimating the arrival time of the audio chirp. As with radio ranging techniques, sound-based ranging methods is subject to significant observation errors due to Non-Line-Of-Sight (NLOS) or multi-path effects, while Pedestrian Dead Reckoning (PDR) based on inertial data have the disadvantage of error accumulation. Both make localization challenging. In this paper, the ranging performance of the Kepler A100 chip is evaluated and experiments are conducted in two typical indoor scenarios to verify the stability and sub-metre accuracy of the ranging results. Meanwhile, this paper implements an indoor positioning system combining audio chirp/Micro-Electro-Mechanical Systems (MEMS)/floor map based on particle filtering algorithm. A new mapping constraint method is applied in the system, which can effectively constrain the particles in a reasonable space. Experiments are conducted in two complex office environments and the results show that the error of the proposed algorithm is about 0.65m in 95% of cases, a reduction of more than 40% compared to only-PDR or only-audio. The effectiveness and stability of the method is demonstrated.