UAVs system for autonomous indoor flight and remote controlled

Nowadays, the industrial environment has strong dependencies on robotic solutions. In fact, these represent a good investment for both major companies and small starts ups. The implementation of robotic and unmanned systems can improve both the efficiency and security of the production. Moreover, in critical and threatening conditions, autonomous vehicles are necessary to limit risks for human operators. Among Unmanned Vehicle System, drones occupy an important and fascinating spot. The greatest feature of this class of vehicle is that, ideally, they can be used in different situations. Drones can provide autonomous flight, both indoor and outdoor, and a great agility also in the tightest places. This master thesis focuses on the development and improvement of the autonomous capabilities of the drone equipped in the FIXIT project. FIXIT is a project carried on by CIM4.0, the Competence Centre for Industry 4.0 of Torino, consisting in the realization of a support system able to move autonomously in an industrial environment and capable of collecting data useful for both in-mission and post-mission analysis. Such system consists in an UAV (Unmanned Aerial Vehicle) that can fly both outdoor and indoor and eventually dock on a rover, which can autonomously map the surroundings and interchange information with the drone. The latter should be able to provide stable flight in GNSS (Global Navigation Satellite System) denied environment exploiting the UWB (Ultra Wide Band) wireless technology and other specific sensors required to perform the flight. The controller is based on ArduPilot, a popular autopilot system, and run on a Pixhawk board. This study aims at selecting and implementing sensors needed to obtain the stability of the aircraft operating in different conditions and on the development of the flight controller to provide better performances and personalized actions. A big challenge with autonomous UAVs in indoor environments is the estimation of an accurate attitude of the aircraft. This is generally provided by the attitude and heading reference system (AHRS). Such system is based on some filtering actions, exploiting the data coming from the sensors on-board. The AHRS runs on the Pixhawk board, and it is realized through a Kalman Filter. Its performances are improved in this work, especially in terms of heading, using an external AHRS completely realized inside the Pozyx master tag which uses an Arduino board as computational unit. The orientation system is based on a complementary filter. In this way the data coming from the ad-hoc calibrated magnetometer are filtered with the ones collected from inertial sensors. Since the magnetometer suffers of high interferences in a small room, an additional heading source is added. This is computed using a system made of two Pozyx tags. This solution is eventually fed to the AHRS filter. The UAV used in this project is an octocopter with coaxial motors, suitable to achieve more stable flight performances. The UWB local positioning is implemented exploiting a sub-library of the ArduPilot’s Extended Kalman Filter. The project is tested in different conditions. First, each new functionality is tested alone in different scenarios. Then, all the contributions are put together, and the UAV behavior is analyzed through an autonomous flight mission inside CIM4.0 laboratories.