Firefighters operate in environments where situational awareness is crucial for safety and effectiveness. This thesis addresses three key challenges: subject localization, hazard recognition and localization, and interaction with challenging environments. Traditional navigation tools face limitations in such scenarios, including GPS-denied environments, compromised radio frequencies, and magnetic interference, making conventional solutions unreliable. In this thesis, we leverage inertial navigation systems (INS) with foot-mounted Inertial Measurement Unit (IMU) systems to overcome personnel localization challenges to then address hazard localization and path planning problems. The main contributions of this thesis includes:• Introduced a time series Support Vector Machine (SVM) forecaster algorithm, classifying 19 distinct pedestrian activities using measurements collected from a foot-mounted Inertial Measurement Unit (IMU), capturing both swing and stance phases of a pedestrian stride. The classification encompasses activities including walking, fast walking, jogging, running, sprinting, walking backward, jogging backward, and sidestepping. The trained time series SVM classifier achieves a 90.04% average classification accuracy across three subjects, resulting in an improvement in navigation accuracy. Integration of motion predictions with the Zero Velocity Update (ZUPT) algorithm significantly reduces navigation errors. The ZUPT-aided INS incorporating the time series SVM forecaster detector outperformed both as standalone INS (improvement by a factor of 197.18×) and a fixed threshold ZUPT-aided INS (enhancement by a factor of 2.85×) in horizontal Circular Error Probable (CEP) navigation accuracy along a short trajectory (on the order of 100 [m]). Additionally, the approach allowed to reduce the vertical drift by factors of 245.34× and 1.16×, respectively.
• Introduced a robust mapping framework utilizing a foot-mounted IMU and a customized handheld device. Achieving precision within 2 [m] for distances less than 20 [m] from the target, the framework demonstrated accuracy in hazard localization. The study identified primary error sources, evaluated contributing factors, and transparently outlined inherent limitations, providing valuable insights for practical applications and future improvements.
• Introduced a real-time guidance system for cooperative coordination and guidance for firefighters in indoor environments. By seamlessly integrating augmented reality (AR) headsets and Windows PC applications, the system provided dynamic and efficient guidance, potentially improving team efficiency and safety. The applications operate in tandem, facilitating real-time tracking of firefighters within a building on a Windows PC and offering visual assistance through an AR headset. The system’s ability to project optimal routes in real-time, guiding teams based on objectives set by fire commanders, ensures effective navigation even in complex scenarios where pre-surveyed maps are unavailable. This approach not only enhances the safety of firefighting missions but also contributes to the overall efficiency of coordination strategies in dynamic indoor environments.