RMIT University develops indoor positioning via smartphone sensing

Getting lost at a shopping mall will be almost impossible with the development of a more accurate and reliable indoor positioning system using smartphone sensors.

Victoria’s Royal Melbourne Institute of Technology (RMIT) University is developing state-of-the-art technologies for indoor positioning and has a prototype system that is demonstrating its great potential for real-life use.

The recent advances of smartphone and sensing technologies has meant indoor positioning, which leverages on smartphone sensing, has become more and more popular in the last few years.

The team has developed an accurate and power-efficient system using magnetometers on smartphones, leveraging on a novel augmented particle filter.

The system, which has received funding from the State Government of Victoria, could be used in shopping malls, airports, train stations, office buildings, hospitals, trade fairs, museums and warehouses.

RMIT University computer science and information technology Associate Professor Tao Gu said the system used the magnetic field sensor data as a fingerprint for each location to identify a user.

The RMIT University team has been working with Melbourne-based company Serraview to benchmark the solution against other indoor positioning systems such as Wi-Fi fingerprinting, Bluetooth fingerprinting and Wi-Fi triangulation.

“The way it works is the user holds a smartphone and downloads a floor map,’’ Professor Gu said. “When they walk with the smartphone, the screen shows the map and also shows their current location.’’

He said many IT companies, including Google and Microsoft, were looking for a good indoor positioning solution.

“It is still an open question, very challenging in terms of accuracy, reliability and scalability.’’

Professor Gu said the system’s uniqueness was its accuracy compared with Wi-Fi.

“A fingerprint you collect may work for one phone, but may not work for the next one because of the signal variation of the Wi-Fi,’’ he said. “The magnetometer data is relatively stable so that can result in better accuracy and also reliability.’’

Professor Gu said the system also had the advantage of being “orientation free’’, which meant users could operate their smartphone as usual at the same time.

The solution on average could achieve accuracy of 1 metre to 2.6 metres.

Professor Gu said the team was still in the process of negotiation for the next stage of the project.

“We think this technology is really promising and we are trying to commercialise it,’’ he said.

RMIT University is also planning to work with another Melbourne company Ecot on an indoor positioning project.

The team has already developed solutions for floor localisation and localisation in metro trains.

“When you have such a long train, it is hard to identify a user at a location,’’ Professor Gu said. “It is critical, for example, if you want to find someone or for emergency services.’’

The research used a sensor fingerprinting approach that was similar to magnetic field fingerprinting.

“We don’t use a magnetometer, but we use a barometer, accelerometer, compass and a gyroscope to identify, to collect the data and to process the data and turn it into a fingerprint and use the fingerprint to localise a user.