Robust Multi-Sensor Fusion for Localization in Hazardous Environments Using Thermal, LiDAR, and GNSS Data.

Area of Science:

• Robotics and Autonomous Systems
• Sensor Fusion and Localization
• Hazardous Environment Navigation

Background:

• Autonomous robots are crucial for search and rescue in hazardous environments like tunnels and disaster areas.
• Sensor failures and errors significantly challenge localization accuracy in these demanding conditions.
• Existing localization methods often struggle with the unreliability of individual sensors in complex terrains.

Purpose of the Study:

• To develop a robust sensor fusion algorithm for reliable autonomous robot localization in hazardous environments.
• To integrate data from thermal cameras, LiDAR, and Global Navigation Satellite Systems (GNSS) for enhanced positioning.
• To ensure continuous localization performance despite individual sensor outages or data compromises.

Main Methods:

• Implemented distinct Simultaneous Localization and Mapping (SLAM) and odometry techniques for thermal and LiDAR sensors.
• Utilized an Extended Kalman Filter (EKF) to fuse data from thermal camera, LiDAR, and GNSS.
• Accommodated varying sensor sampling rates and simulated single-sensor outages during field testing.

Main Results:

• The proposed sensor fusion algorithm demonstrated reliable localization performance in challenging urban environments.
• The system effectively compensated for individual sensor failures, maintaining localization accuracy.
• Field tests validated the algorithm's robustness under simulated real-world conditions with sensor outages.

Conclusions:

• Robust sensor fusion is essential for dependable autonomous navigation in hazardous and unpredictable settings.
• Integrating thermal, LiDAR, and GNSS data via an EKF provides a resilient localization solution.
• The developed algorithm offers a promising approach for enhancing the safety and effectiveness of autonomous robots in critical operations.