Smart Organizations of Unmanned Aerial Vehicles Using SOMs for Monitoring Cars on Urban Roads

In the Cooperative Multi-Robot Observation of Multiple Moving Targets (CMOMMT) problem, a team of autonomous robots with equivalent sensing and motion capabilities in a delimited 2D area cooperate to perform the observation of moving targets, aiming to maximize the time during which at least one observer is observing each target. More recently, some control strategies have been proposed for the problem focusing on targets on urban roads. This work advances and proposes the CMOMMT-URBAN3D problem, in which unmanned aerial vehicles (UAVs) observe targets that are automobile vehicles on urban roads containing 3D obstacles, buildings, and other constructions of different sizes. Our approach presents a new graph formalism for UAV agent organizations, a new concept called a mobile environmental pattern, and a new behavior using self-organizing neural networks (SOMs) to conceive these strategies. Performance evaluations of the UAV team demonstrated a tendency of superiority, in some scenarios, of strategies that consider our new formalism and your theoretical results as well as our new behavior in the project of the organization structure.