Tracking emergency responders in challenging environments

First responders, such as firefighters, often operate in extreme environments and may sometimes become disoriented (e.g., inside buildings) and disabled even in relatively small incidents (see Figure 1). Many incidents (such as the events of 9/11 and the Worcester Cold Storage Warehouse fire that killed six firefighters) have highlighted the need for an accountability system to accurately locate, track, monitor, and visualize the locations of responders on a geospatial map. This will allow incident commanders and tactical-decision makers to virtually observe personnel movements in real time to avoid loss of life.

Figure 1. Complex incident environment.

Currently, global-positioning systems (GPS) are good in the open field. However, tracking personnel in three dimensions and inside complex buildings (where GPS signals are nonexistent) is difficult. Many radio-frequency (RF)-based systems rely on triangulation and time-difference-of-arrival techniques to compute positions. Other systems, many still in early development, use components such as inertial measurement units (IMUs) coupled with RF techniques to improve position accuracy. Others have experimented with existing (television, radio, or satellite) signals. Many of these technologies are either still in development or not yet robust for field deployment. Lack of a suitable technology and the stringent requirements¹ of the user community have made the problem even more complex, so that systems that provide adequate performance are heavy, cumbersome to use, and too expensive.

In response to these challenges, we have started development of the geospatial location accountability and navigation system for emergency responders (GLANSER). The system is a ‘cocktail solution’ in which several components have been fused together to provide an estimate of the user's location, whether inside or outside a building. We have combined GPS, IMU, ultrawide-band ranging radio, Doppler radar, as well as a magnetometer, compass, pedometer, and altimeter, to fit into a 2×4×6in³ wearable electronic unit. This combination of sensors works in harmony so that when GPS is not available, or in periods of suboptimal RF ranging, other signals are exploited. Aided by the onboard Kalman filter, all measurements are processed to compute the responder's location to submeter accuracy. GLANSER's mobile ad hoc mesh network is then used to continuously transmit this information to a base station at the command post.

One of the architectures that we are considering for implementation includes responder-host locator nodes that blend multipath-mitigated assisted-GPS and cross-range measurements derived from a wireless-mesh network of other hosts, as well as inertial measurements to form an estimate of the host's 3D position, orientation, and trajectory. Vehicle-installed base-station nodes similarly participate as peers in the mesh network, helping to enhance host-positioning performance. Base nodes monitor the network of host nodes and provide additional corrections for GPS atmospheric-propagation errors and any inertial sensor drift. Operation requires no setup time other than turning the nodes on. Signal processing effectively mitigates multipath reflections for both peer-to-peer crosslink and GPS range measurements. We use other techniques to eliminate range-measurement biases caused by problematic short-delay multipaths. These biases often exceed tens of meters in urban and indoor settings, and can create equally large navigation errors. In addition, autonomous map-generation and map-matching algorithms help constrain IMU drift by automatically determining the location and orientation of building features and applying the appropriate corrections. During periods of suboptimal RF ranging, IMU provides the only means for localization. By reducing the magnitude of IMU errors, this technology enables use of low-cost IMUs.

To date, development of GLANSER has made significant progress. We tested our prototypes (see Figure 2) in several venues and demonstrated (see Figure 3) that an accuracy of three meters in all dimensions is achievable. The altimeter performance in several test cases was very good (see Figure 4) and proved that, in simulated incident conditions, the altitude could be computed to an average accuracy of three meters. While the prototypes were still large, GLANSER has proved that the cocktail solution is a correct approach and allows for integration of future components as they are developed.

Figure 2. Prototype geospatial location accountability and navigation system for emergency responders.

Figure 3. Test case. Colors represent tracks for each person wearing the tracking unit during the tests. The tracks (starting from the green area) vary because users walked slightly different paths. Once inside the building, they crossed a hallway and converged in the (red) area where the simulated downed responder was hidden.

Figure 4. Altitude performance. The red line represents the real floor height, while the jagged blue line shows the sensor readings. These oscillate because the measurements are approximate. The sharp vertical lines correspond to data dropouts. MSL: Mean sea level.

Our prototype system helped us learn the intricacies of location tracking but also identified some challenges. First and most importantly, we know that end-user involvement is necessary and drives design features, specifically cost, size, weight, automation, power, and maintenance. Several other technology challenges still remain, including multipath, loss of signal, and error correction that result in degradation of accuracy. Current concepts/systems work some but not all of the time, so reliability and robustness is important. Simple integration with existing equipment (such as breathing apparatus, personal alert safety system, and personal protection equipment) is of great importance and a driver for system interfaces. Cost is a major consideration that also drives component design and choice.

Indoor tracking of personnel is still challenging. We are currently developing a solution to enable such a capability. GLANSER is on track to field-test two similar but competing architectures in the summer of 2010. Upon successful demonstration of performance and robust operation, initial production units will be spun off so they can be field-tested with first responders.