Coordinated Search and Mapping Using Swarm Robots

Abstract

Natural disasters can cause widespread devastation, leaving survivors trapped in rubble or buried under debris. The time it takes to identify and rescue survivors is critical, this research seeks to demonstrate how Cooperative behaviors in multi-robot systems can significantly improve the speed and accuracy of victim identification following a natural disaster. This research focuses on the navigation of the robot swarm, mapping, and victim identification. The navigation approach is determined by the particle swarm theory, with the robots serving as the swarm’s agents. It moves to a destination location by avoiding impediments, maintaining group compactness, and using the attraction and repulsion forces typical of swarm particle systems. Here, each agent independently uses YOLO 5 to find the victims. When the agents agree that a victim may be present, a robot is chosen based on the Election by Consensus protocol. The chosen robot then moves around the area and maps it using RTAB, which generates a 3D point cloud and a 2D occupancy grid map of the area.