Embedded Model Control for UAVs: theoretical aspects, simulations and experimental results

Unmanned Aerial Vehicles (UAVs) and, more specifically, n-copters have come to prominence in the last decade. Indeed, unmanned vehicles may have several applications in society, spanning from complex operations, also in potentially hazardous environments for humans to more entertaining purposes. Furthermore, UAVs have drawn great attention in the automatic control research community. This is mainly due to two reasons. First of all, designing a control for this non-linear and underactuated system can represent a stimulating challenge for control researchers. Secondly, n-copters, being typically mechanically simple and fast-prototyping devices, are widely considered as a good technology for testing a wide range of control algorithms and designs, also employing a wide range of sensors. In fact, the possibility of having several low cost sensors on-board enables the implementation of many navigation solutions as well as sensor calibration algorithms. In this work, the control unit design for a quadrotor was addressed. In particular, the study regards the Borea quadrotor which is part of an internal project (Borea) of the former Space and Precision Automatics research, now Systems and Data Science group, at Politecnico di Torino. The Borea project aims to test Guidance, Navigation and Control (GNC) algorithms designed within the framework of the Embedded Model Control (EMC) methodology. In particular, one of the main project objectives regards the testing of planetary landing algorithms because of the similitude in the command authority between n-copters and spacecrafts during the landing phase. In fact, both n-copters and spacecrafts can provides a thrust vector which is constant in direction whereas its intensity can be regulated. The EMC framework matches the GNC architecture perfectly. More important, EMC is a methodology based on an internal model which includes the uncertainties, in the form of disturbances, that have to be rejected. Indeed, the main design effort is focused on the internal model design, which is the core of the whole control unit. The Borea UAV has been endowed with a control system in order to control its position, velocity, and attitude. These results has been achieved by means of a well structured design process which started from the plant modelling and arrived to the flight test. Indeed, the process has involved intensive numerical simulation and control refinements as well as multi-staged tests and model validations. During the design process some neglected dynamics has turned out to be very important for the control design and their identification was revealed mandatory. On the other hand, the control problem was separated into two independent controllers in order to have a more simple controller which makes the quadrotor able to fly allowing to test all the subsystems, improve the simulator fidelity and support the design and validation of the second controller. Each controller has required specifics flight tests aimed to validates particular functionalities and control performance. Testing has included the design and building of a single axis test-bench in order to perform the very first control tuning in a safety way. The objective of the first controller was the attitude stabilization of the quadrotor in order to perform a hovering flight initially and the attitude tracking later. The design of the attitude controller has required the identification of the actuator dynamics as well as the sensors calibration. The attitude control unit has been implemented in all its parts and successfully tested in real flight. As mentioned before, the next step has been focused on controlling the quadrotor position within a limited flight area. In particular, this study investigates the use of the feedback linearization approach as a novel way to design the internal model for EMC. The feedback linearization allows us to collect all the non-linearities at the command level. EMC, by means of a disturbance dynamics model, makes possible to estimate and then reject the non-linear terms through the control law. The control solution has been validated by means of intensive numerical simulations and real-flight tests. As a final result, the control units developed in this work enhance the EMC applicability to non-linear systems, such as quadrotor UAVs, and evidence the EMC disturbance rejection capabilities.