ROS/Gazebo based simulation environment for UAV controller performance evaluation

This project aims to create a ROS/Gazebo-based environment for UAV controller performance evaluation through simulated experiments. A general purpose interface is also developed to integrate MATLAB/Simulink with ROS/Gazebo. The combined setting allows for incorporating realistic environment constraints and uncertainty, system model requirements, and other exogenous factors into the testing framework. This work takes as case of study a quadrotor, which is widely used nowadays for military, commercial and private purposes. These systems have a highly nonlinear behavior and different challenges to deal with; like sensor noise, external disturbances (i.e. wind gust), hard and precise maneuvers and payload release. The model of the quadrotor is implemented in Gazebo taking into account a detailed aerodynamics model and feedback sensor noise. All these effects are implemented in the URDF model with plugins that are built on the testbench and can be activated by the user depending on the simulation purpose. The use of ROS/Gazebo and the implementation with MATLAB is not a straightforward application, so this thesis try to create and explain a better and clear workflow to achieve the controller simulation. The work presents the experimental data from simulation of different types of control techniques to highlight the reliability of the interface between the software tools. The simulated experiments on ROS/Gazebo and their performance evaluation lead to a better testing phase and an optimization in the software-in-the-loop step in the design flow.