Working towards scenario-based evaluations of first responder positioning systems
2011 (English)Conference paper (Refereed)
It is indeed a challenge to create a highly-accurate positioning system for GPS-denied environments, such as indoors and in dense urban environments, or during harsh electromagnetic interference conditions. A priori information concerning for instance building layouts, magnetic field or image information obtained through extensive pre-surveying, or pre-installed infrastructure (e.g. RFID-units or WiFi base-stations), will not be available in many scenarios. However, a reliable positioning system with seamless outdoor and indoor coverage would increase the safety of first responders significantly. Such a system should be lightweight, small, power efficient and relatively inexpensive, yet still provide a high accuracy during extended indoor operations. A camera-based reference system which provides reliable ground truth data in relatively large indoor environments, using a minimum of pre-installed infrastructure, has been designed and evaluated. The reference system will be used to evaluate how errors in for instance foot-mounted inertial navigation systems (INS) or pedestrian dead-reckoning systems grow over time. Loop-closure error analysis is not sufficient for the evaluations of INS since the shape of the trajectory influences the position error. A low-cost, lightweight reference system is often needed when attempting to evaluate the performance of high-accuracy positioning systems during scenario-based exercises. In this work ARToolKitPlus have been used to estimate the camera position relative to a set of pre-installed visual markers. Experiments performed in a very accurate VICON lab show that the toolkit provides sufficient localization accuracy for the intended accuracy. The error is normally less than 0.1 meters at a distance of 1.75 meters from the marker. When the camera-based reference system is used to evaluate a foot-mounted INS, a number of checkpoints with markers will be passed several times. This way the error growth rate in the IMU-based system can be evaluated. Preliminary results from measurements performed with foot-mounted INS are provided and compared with the reference system. Practical considerations concerning sensor fusion of foot-mounted INS with additional sensors, such as magnetometers, barometric altimeters and GPS, will also be provided and particular needs for continued research are pointed out.
Place, publisher, year, edition, pages
IdentifiersURN: urn:nbn:se:kth:diva-85900OAI: oai:DiVA.org:kth-85900DiVA: diva2:500168
Precision Indoor Personnel Location and Tracking Technology Workshop, Worcester, 1-Aug-2011
QC 201204112012-02-132012-02-132012-04-11Bibliographically approved