A New Sensor-based Method for Localizing a Person Indoors
In his dissertation, Jari Saarinen, M.Sc., has developed new solutions for a major problem in localization: i.e., how to localize a person indoors?
Earlier, localizing a person indoors was possible only if systems, such as localization systems based on WLAN, had been installed and calibrated for the building beforehand. In the method of Saarinen these are not needed. The solution lies in different sensor combinations, which the person that we want to localize carries with him/her. Being able to localize a person indoors makes it easier and safer, for example, to coordinate fire and rescue operations.
"The person we want to localize carries with him/her sensors, which collect data, among other things, about the steps the person has taken, acceleration and turning speed. By combining the data generated by the different sensors, the person can be localized on the base map of the building. There is no need to install anything in the building itself," Jari Saarinen puts it in simpler terms.
Both dead reckoning with inertial sensors as well as laser based dead reckoning feature as important research topics for the dissertation work. Dead reckoning is based on angle estimation using a compass and a gyroscope as well as on estimation of step length. The work presents two alternative methods for estimating step length: one of these methods uses ultrasound and the other one accelerometers.These are supplemented with a laser based method: by comparing the distance images, in succession, that are produced by a laser range finder the displacement between the images can be found. Final location information is obtained by fitting the data on the map. The work presents three different variations of map based localization. These are based on the well-known Monte Carlo localization method.
The methods were tested, among others, in a simulated fire and rescue operation, in which a team composed of humans and robots had a joint task to perform. The focus of Saarinen's dissertation was on indoor localization on a single level. Three-dimensional localization was dealt with, preliminarily, based on the same methods.The research will continue if funding can be ensured.
"In future, it is possible to install localization sensors wherever needed, even in mobile phones. With the help of a sensor-based personal navigation system one could, for example, easily locate a sought-after product in a supermarket, mine safety could be increased by knowing the location of each worker, and in hospitals one could locate the doctor to help in a case."
Jari Saarinen, M.Sc., will publicly defend his dissertation "A Sensor-Based Personal Navigation System and Its Application for Incorporating Humans Into a Human-Robot Team", dealing with the field of automation technology, on Thursday, 25th June 2009.
Source: Academy of Finland
