[Paper Review] A Novel Indoor Positioning System for unprepared firefighting scenarios

Vamsi Karthik Vadlamani, Manish Bhattarai+2 more|arXiv (Cornell University)|Aug 4, 2020

Indoor and Outdoor Localization Technologies12 references4 citations

TL;DR

This paper proposes a novel indoor positioning system (IPS) for unprepared firefighting scenarios using optical flow-based video compass for orientation estimation and IMU-based activity recognition to mitigate drift. By fusing lidar, IMU, and thermal video data with support vector regression and thresholding, the system achieves ~2-meter accuracy in unknown, dynamic environments without prior infrastructure, enabling real-time situational awareness for first responders.

ABSTRACT

Situational awareness and Indoor location tracking for firefighters is one of the tasks with paramount importance in search and rescue operations. For Indoor Positioning systems (IPS), GPS is not the best possible solution. There are few other techniques like dead reckoning, Wifi and bluetooth based triangulation, Structure from Motion (SFM) based scene reconstruction for Indoor positioning system. However due to high temperatures, the rapidly changing environment of fires, and low parallax in the thermal images, these techniques are not suitable for relaying the necessary information in a fire fighting environment needed to increase situational awareness in real time. In fire fighting environments, thermal imaging cameras are used due to smoke and low visibility hence obtaining relative orientation from the vanishing point estimation is very difficult. The following technique that is the content of this research implements a novel optical flow based video compass for orientation estimation and fused IMU data based activity recognition for IPS. This technique helps first responders to go into unprepared, unknown environments and still maintain situational awareness like the orientation and, position of the victim fire fighters.

Motivation & Objective

• To address the lack of reliable indoor positioning in unprepared, dynamic firefighting environments where GPS and RF-based systems fail.
• To overcome limitations of existing IPS techniques—such as Bluetooth beacons, WiFi fingerprinting, and SFM—due to high temperatures, environmental instability, and lack of parallax in thermal imagery.
• To enable real-time, drift-resistant position and orientation tracking for firefighters using only thermal video, IMU, and lidar data.
• To develop a system that functions without prior knowledge of the environment, such as beacon or access point placement.
• To enhance situational awareness by accurately estimating orientation and position in real time, even in heavy smoke and extreme heat.

Proposed method

• Uses thermal video from an IR MSA 5200HD2TIC camera (320x240 resolution, 30 fps) to estimate orientation via optical flow-based video compass, bypassing vanishing point estimation that fails in low-edge thermal images.
• Applies IMU-based activity recognition to detect turns and reset position estimation, reducing drift from inertial integration.
• Fuses accelerometer and lidar data using least squares estimation with initial condition estimation to correct for cumulative errors during turns.
• Employs Support Vector Regression (SVR) with a hybrid cost function (Vapnik’s ε and Huber robust loss) to model lidar distance data and improve generalization under sparse, noisy samples.
• Applies differentiation to SVR-estimated position to compute velocity, followed by thresholding using median filtering to enforce physical plausibility and remove outliers.
• Uses sigmoid convolution and peak detection on orientation data to identify turns and trigger reinitialization of position estimation algorithms.

Experimental results

Research questions

• RQ1 Can a video compass based on optical flow provide reliable orientation estimation in low-parallax, low-contrast thermal images typical of firefighting environments?
• RQ2 How can IMU-based activity recognition be used to detect turns and reset position estimation to mitigate inertial drift in unprepared indoor scenarios?
• RQ3 To what extent can SVR with robust cost functions improve position estimation accuracy when fusing sparse and noisy lidar and IMU data in dynamic, high-temperature environments?
• RQ4 Can thresholding and median filtering of velocity estimates enforce physical consistency and reduce artifacts in position tracking without prior environmental maps?
• RQ5 How does the fusion of optical flow, IMU, and lidar data compare to existing IPS methods in terms of accuracy and robustness in unprepared, unknown indoor settings?

Key findings

• The proposed IPS achieves a path reconstruction accuracy of approximately ±2 meters in unprepared indoor environments, even without prior knowledge of the building layout or beacon placement.
• The system successfully detects turns using orientation data and sigmoid-based peak detection, enabling reinitialization of position estimation and significantly reducing cumulative drift.
• Support Vector Regression (SVR) with a robust cost function effectively models lidar distance data, showing excellent generalization and resilience to outliers despite limited training samples.
• Thresholding and median filtering of velocity estimates eliminate physically implausible spikes, improving the stability and realism of the estimated trajectory.
• The optical flow-based video compass outperforms vanishing point estimation in thermal imagery, where edge detection fails due to low resolution and dynamic range.
• The system demonstrates robustness in real-world firefighting conditions, including heavy smoke and high temperatures, where GPS and RF-based systems fail.