An Infrastructure-Free Localization System for Firefighters

Carnegie Mellon University

---

The team is developing a rapidly deployable infrastructure-free localization system to track first-responders inside potentially harsh environments. The goal is to provide team members outside of the facility (like fire safety chiefs) a live feed on a tablet or computer a read-out with the position of each crew member. Given the hostile nature of burning structures and the time criticality of missions, this requires that a system can track firefighters without any pre-installed internal and limited external infrastructure, and without assuming knowledge of the structure’s layout. For a system to be practically adopted at scale, it also needs to be low-cost and extremely simple to configure and deploy. -July 2019

Meet the Team

Project Overview

This project focuses on the development of a rapidly deployable infrastructure-free localization system to track firefighters inside of a structure such as a building. The goal of the project is to provide fire safety chiefs who are responsible for team accountability a live feed on a tablet or computer outside of the facility that can show the position of each firefighter within. Given the hostile nature of burning structures and the time criticality of missions, this requires that a system can track firefighters without any pre-installed internal and limited external infrastructure, and without assuming knowledge of the structure’s layout. For a system to be practically adopted at scale, it also needs to be low-cost and extremely simple to configure and deploy. Such a system can be created through the integration of three maturing technologies:

1. compact and accurate Microelectromechanical Systems (MEMs) inertial measurement units (IMUs)
2. accurate time-of-flight Ultra-Wide Band (UWB) ranging radios and
3. compact long-range wide-area network (LR-WAN) digital radios.

No single technology would work in isolation. We envision a system consisting of a small number of external beacons (3-4) that are deployed around the perimeter near ingress points and a small wearable unit attached to each firefighter's belt or air pack. Each external beacon would have a GPS receiver, sub-GHz LoRa radio, UWB radio and air pressure sensors. The wearable unit on each firefighter would have similar hardware with the addition of an IMU.

The external beacons would determine their locations using GPS to initialize firefighter locations as they enter the structure using precise (˜10cm) UWB ranging. The beacons would have one receiver at ground-level and could telescope tens of meters above ground with another receiver to provide vertical diversity for multilateration and air pressure sensing at multiple points to estimate vertical height. When firefighters enter the facility (always in pairs of two) each wearable unit would use its IMU to track their movement from ingress points and their orientation over time. The UWB radios would measure inter-firefighter and beacon ranges whenever possible.

Using a mobile network localization approach inspired by work in Simultaneous Localization and Mapping (SLAM) the system would be able to compensate for IMU drift by using inter-device ranging from the UWB radios. All data would be streamed out of the structure using the LoRa radios to an interface available to safety chiefs. It is critical that the system provide both position and bearing such that a firefighter in zero visibility can be given relative radio commands instructing them to ingress, egress or turn left/right. Once a robust core localization system is in place, there are multiple enhancements that can be added to the command software that shows team locations and suggest paths based on previous routes.

The team has designed and built an acoustic and UWB-based localization system and won the 2015 Microsoft Indoor Localization competition with an average accuracy of 31cm. Through a longstanding collaboration with the Bosch Research the team is currently piloting an indoor localization system for smart phones in the Pittsburgh Convention Center. Many of the algorithms we are developing for auto-configuration of our system can be applied to the problem of mobile network localization. As part of this project, the team has interviewed local firefighters to elicit requirements and are in discussions about how best to benchmark the system during training scenarios.

<< Back to the PSIAP 2017 - Location-Based Services Page