AbstractFigures from this paperAbstractFigures from this paperAbstractFigures from this paperAbstractFigures from this paper

This study proposes a behavior tree (BT)-based autonomous task management system for tunnel firefighting robots, integrated with ROS2 and Autoware frame- work. The system coordinates perception, navigation, and fire-extinguishing modules by leveraging BT’s modularity and responsiveness. A visibility-reward- augmented Hybrid A* algorithm ensures flame visibility during navigation, while flame detection and ROS2-enabled localization guide water-cannon con- trol. Virtual and real-world experiments demonstrate effectiveness: navigation achieved 20.22 s average time with 6.21° angular deviation, and water-cannon extinguishing averaged 1.77 to 3.87 s per flame across angles. Results con- firm robust task execution in dynamic tunnel environments, highlighting the framework’s adaptability for firefighting scenarios.