Development of a simulation environment for precision agriculture applications with Unmanned Aircraft Systems

In this work, the Open Source technologies PX4, ROS2 and GazeboSim have been integrated to provide the Department of Mechanical and Aerospace Engineering of the Politecnico di Torino a test environment to carry out their research activities. In particular, this environment allows simulation in the loop testing for the control system of an agricultural spraying drone. The first part of this work focused on configuring the GazeboSim world's terrain in a computationally efficient but detailed description, based on an aerial laser scansion of a vineyard field. Then the GazeboSim drone's model has been configured, in particular, other than the essential flight sensors, a depth camera was added and the data stream was made available for ROS2 nodes. The second part of the work was based on connecting the PX4 autopilot with the sensors simulated in GazeboSim in the most modular way possible. Unfortunately, here the work was limited by a missing functionality in PX4, which did not allow the connection of an externally simulated global navigation satellite system. Anyway, by slightly reducing the modularity of the simulation environment, the issue was solved. The last part of the work was dedicated to the setup of the ROS2 main package with: a hierarchy of launch files; the definition of the robot topology, needed for the depth camera's pointcloud interpretation; a simple controller node, setting the references for the autopilot. Then an additional package has been developed to provide the GazeboSim model with two custom plugins. The first plugin enables the control of the simulation wind speed, using a GazeboSim topic and regulates the application of a drag force on the drone frame. The second plugin models the variable mass of the spraying system's reservoir and all the gravitational and inertial forces related to it. It also predisposes the modeling of the sloshing effect force, whose description is outside the scope of this work. Also, this second plugin controls a graphical visualization of the spraying activity of the drone. Finally, an additional ROS2 package has been developed with the aim of driving the wind topic taken in input by the first plugin. In conclusion, a campaign of simulations was carried out to evaluate the performances of the simulator and compare them with the expected results.