FirebotSLAM: Thermal SLAM to Increase Situational Awareness in Smoke-Filled Environments

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Extreme environments like fires severely limit visibility for firefighters.
• Unmanned Ground Vehicles (UGVs) can enhance situational awareness in hazardous conditions.
• Existing Simultaneous Localization and Mapping (SLAM) methods struggle with thermal image data.

Purpose of the Study:

• To develop the first algorithm for robot odometry and 3D mapping using only thermal images.
• To improve firefighter safety in fire incidents through enhanced robotic perception.
• To address the challenge of robust feature extraction in thermal imagery for SLAM.

Main Methods:

• Proposed FirebotSLAM, a novel algorithm for thermal SLAM.
• Benchmarked feature extraction and description methods on fire incident datasets.
• Integrated the best feature combination into a state-of-the-art visual SLAM algorithm.

Main Results:

• FirebotSLAM computes robot odometry and generates 3D maps from thermal images.
• Achieved global trajectory optimization via loop closure detection.
• Demonstrated the first successful test of a thermal SLAM algorithm in a real fire environment.

Conclusions:

• FirebotSLAM enables robots to navigate and map hazardous, low-visibility environments using thermal data.
• The developed algorithm significantly enhances situational awareness for emergency responders.
• This technology validates the operational applicability of thermal SLAM in real-world fire scenarios.