Enhanced Pedestrian Dead Reckoning Sensor Fusion for Firefighting

Knowing the exact position of firefighters in a building during an indoor firefighting operation is critical to improving the efficiency and safety of firefighting. Since GPS lacks the required accuracy in indoor environments, an alternative solution is demanded. Examples are radio or Wi-Fi triangulation or magnetic field mapping. However, first responders' unique challenges call for an approach only relying on body-worn sensors. This is due to the fact, that triangulation or mapping is not available in every building prone to fire. The so-called Pedestrian Dead Reckoning (PDR) approach, estimates an individual's position relative to a starting point. PDR solutions utilize step-detection as their main means of position tracking. While step-detection produces accurate results during normal walking processes, during other dynamic activities like crouching or running the accuracy is significantly reduced. In this paper we propose an approach that features an enhanced sensor data fusion algorithm to increase position estimation accuracy in various moving scenarios. The enhanced algorithm fuses position data from an inertial measurement unit based step-detection with tracking camera position and velocity data in an extended Kalman filter. To evaluate the quality of the enhanced sensor fusion algorithm, results from a verification campaign in a camera-based, high precision measurement environment are presented. This environment allows a sub centimeter tracking resolution of an individuals position. With the enhanced sensor fusion, a mean error of less than one meter is achieved which is significantly lower than using step detection only and thereby provides adequate tracking performance of indoor firefighting personnel.