Design and test of a multi-agent robotic system for inspection: coordination and integration of a drone and a robotic rover.

The purpose of this project is the development of a multi-agent system capable of performing aerial and ground inspections. The possibility of having two vehicles inspecting from two different areas, one from the ground the other from above, guarantees a more accurate inspection in places that are not easily accessible to humans, for example, limited spaces or too high areas. Another advantage is the dynamic inspection which, compared to fixed cameras, allows the elimination of the dark areas of the latter. The types of inspection for which this system has been designed are search and rescue and industrial inspection. During search and rescue operations, for example, it is important to minimize the potential dangers that can endanger both the life of the research groups and the people to be rescued (unstable soils or free electric cables). In industrial inspections, on the other hand, it can be useful as a surveillance system in large areas, for example, container areas, in large warehouses as product control or to inspect dangerous areas. The two tasks that have been developed, during thesis work, are described here: the autonomous flying capabilities of a drone and the development of a research rover able to be transported on by the drone through a gripping system, developed to ensure minimum bulk and weight. The development of a predefined path that the drone will follow for the inspection flight, the research and the precision landing on a visual marker.Within this project, the visual marker is positioned above the rover so that the drone can pick it up and bring it back. DJI Matrice 100 quadcopter has been used as development platform, along with a single-board computer, which acquires, processes and exchanges data between drone and rover. The rover has been designed and integrated through a WiFi Development Board that receives data from the main computer and sends the commands to the motors and camera installed inside. To build and run codes across the multiple computers used in this project, ROS (Robot Operating System) has been used to provide tools and libraries. The system has been first tested in simulation through the official application of the drone and GAZEBO (an indoor and outdoor simulation environment). Later, experimental tests have been performed on each task of the project, and then concluded with a complete simulation of the entire mission.