Real-Time Fire and Smoke Detection for Trajectory Planning and Navigation of a Mobile Robot

Abstract

Mobile robots have many industrial applications, including security, food service, and fire safety. Detecting smoke and fire quickly for early warning and monitoring is crucial in every industrial safety system. In this paper, a method for early smoke and fire detection using mobile robots equipped with cameras is presented. The method employs artificial intelligence for trajectory planning and navigation, and focus is given to detection and localization techniques for mobile robot navigation. A model of a mobile robot with Omni wheels and a modified YOLOv5 algorithm for fire and smoke detection is also introduced, which is integrated into the control system. This research addresses the issue of distinct objects of the same class by assigning each object a unique identification. The implementation not only detects fire and smoke but also identifies the position of objects in three-dimensional space, allowing the robot to map its environment incrementally for mobile navigation. The experimental results demonstrate the high accuracy achieved by the proposed method in identifying smoke and fire.

Research and Development of a Traffic Sign Recognition Module in Vietnam

“…The head is the final component responsible for object recognition based on the extracted features. Some studies have successfully used Yolo deep learning models to solve real-time application problems, such as smoke detection for trajectory planning and navigation of a mobile robot [13] and recognition of road surface anomalies [14].…”