A Machine Vision Based Fire Fighting Robot with Real-Time Fire Localization

---

Abstract

This paper presents an analysis of deep learning algorithms, particularly YOLOv8, into an intelligent robotic system aimed at detecting and pinpointing fires in camera imagery. Our method has achieved a noteworthy 94.5% accuracy in classifying fire detection, representing a significant achievement in real-time object recognition. While recognizing this achievement, the research acknowledges significant opportunities for further enhancement in the detection of fire types. The implementation of machine vision in the advanced robot establishes the ground-work for adaptability in response to different fire situations. This initial milestone signifies a promising advancement in the direction of more efficient fire response, and also highlights the ongoing progress in the convergence of deep learning and robotics. The results suggest practical uses for intelligent robots in real-life scenarios, particularly in emergency response and public safety situations.

---

Keywords

You Only Look Once (YOLO); Global System for Mobile Communication (GSM); Integrated Development Kit (IDE); Simultaneous Localization and Mapping (SLAM).