Current positioning infrastructures such as Global Positioning System (GPS) are great for outdoor localisation but are limited and most of the time unavailable in indoor applications. Current smartphones provide new opportunities for user indoor localisation by leveraging low cost embedded sensors. This paper presents an indoor positioning system using accelerometers, gyroscopes and magnetometers which are readily available in most current smartphone. The method proposed is a practical solution for smartphone-based localisation that could minimise the errors due to noise from the low cost inertial sensors and random handling condition of the smartphone by the user. The main focus in increasing the accuracy of indoor localisation system in this paper is by correcting the path taken by the user by implementing a map-based particle filter that takes into account situations where all the particles are at an invalid position and considered dead. This localisation system assumes that the map provided is complete hence positions and paths that lie outside of a valid space are considered impossible for the user to be in and invalid. Experiments have been conducted to test the performance of the proposed method. Two different Android smartphones were used and 30 samples were collected with each sample covering a distance of more than 100 metres. Results from experiments show that the proposed method was able to localise a person in an indoor environment with a mean error of less than 2 metres when the final position is compared to the real final position.
展开▼
