During a rescue, firefighters need to navigate through dangerous environments with potentially very limited visibility due to smoke. Satellite-based position technology (GNSS) allows rescue workers to operate efficiently in many outdoor emergency scenarios, but indoors the GNSS signals penetrate insufficiently, making localization inaccurate, unreliable or in most cases completely unavailable.

We address this problem by employing pedestrian dead reckoning (PDR) to track the position of the firefighter’s boot, in which we explicitly take into account the special gait patterns of a firefighter during deployment.

The firefighters are equipped with ProMove-mini IMU sensor modules on each boot. Each sensor module contains an ProMove-mini device that measures acceleration and angular velocity, a compass sensor that measures the local magnetic field vector and a barometric sensor that measures the absolute air pressure. These signals are processed into a position estimate for each sensor module individually. This is performed by integrating the ProMove-mini signals into a displacement and orientation estimate. The current position estimate is determined from the starting position and the current estimate for the displacement relative to that starting position. The compass and barometric sensors are used to reduce the errors in the heading and altitude estimates respectively. Data is collected from several brigade members during exercise, stored on each ProMove-mini IMU module and collected over a 2.4 GHz radio link by a gateway device, which is controlled by a laptop, as shown in the figure below.

Overview of the firefighter system

The figure below provides a schematic overview of the data processing involved in our system. The top part shows the sensors contained in our foot sensor module: two accelerometers, a gyroscope, a compass sensor (magnetometer) and pressure sensor (barometer). The rest of the figure shows how the signals from these sensors are processed into a three-dimensional signal representing an estimate of the foot displacement.

Overview of data processing steps

The system is tested in the test building of Twente Safety Campus (TSC). The building is normally used for the training of firefighters and has facilities for initiating controlled fires. The trials follow as close as possible the procedure of real interventions, in the way the firefighters move around the affected area and inside the building. They first make a round of the building for site recognition. Subsequently, as they enter the building, they move along the walls with the “firefighter step” to explore the inside landscape. To test the consistency of measurements among different sensor nodes, we equip each boot of the firefighter with two ProMove-mini IMUs, such that there are in total four sensor nodes that follow the track. ProMove-mini no. 127 and 128 are on the left-foot and no. 9 and 129 are on the right-foot.

The figure below provides a detailed view of the track of ProMove-mini 127. We can see the path surrounding the building, entering the door and exploring the ground floor rooms. This IMU is placed on the left foot and suffers less from hitting and stamping. The accuracy is quite good and the resulting track follows the real path well.

Results for IMU 127 on the left foot

The figure below shows the computed PDR tracks for all four ProMove-mini sensor nodes. We notice the clear difference between the left-foot IMUs (127 and 128) and the right-foot IMUs (9 and 129). The firefighter hits the walls and stamps the floor much more with the right foot than with the left foot and this is shown clearly in the better accuracy of ProMove-mini 127 and 128 placed on the left foot compared with 9 and 129, which are placed on the right foot.

Results for all IMUs (127 and 128 on left-foot, 9 and 129 on right-foot)

More information about this work can be found in the PhD thesis Pushing the limits of inertial motion sensing by Stephan Bosch.

This work is performed as part of the AIOSAT (Autonomous Indoor & Outdoor Safety Tracking System) European project, which aims to provide solutions for safely tracking firefighters in both outdoor and indoor environments.