Autonomous navigation of non-perception firefighting robot through CCTV-informed vision sharing

Abstract

Deploying smart firefighting robots indoors is a prospective measure to realize fast-response and zero-casualty firefighting operations. Most existing firefighting robots require manual control or pre-input information for navigation. This work proposes a CCTV-informed autonomous navigation system for robot to reach target flame via the safest and fastest route. This proposed system uses the vision of building CCTV-camera network and provides a full-process navigation pipeline to guide imperceptive firefighting robots. Demonstrations show that the navigation system enables the robot to have a complete perception of floorplan and understand the evolution of fire scenes. The robot can autonomously move towards the target flame even if carry-on vision sensors are blocked or damaged. Finally, the dynamic system responsiveness is validated to prove its effectiveness in addressing randomly moving obstacles. This framework can be further integrated with smart building maintenance system, which offers low-cost, early-response, and more resilient solutions for firefighting and safety patrol robots.