Master thesis presentation Björn Lönnquist

Examiner: Petter Falkman
Supervisor: Lars-Berno Fredriksson
Opponents: Simon Hansson and Viktor Fredholm

Abtract
Control systems for automotive, aerospace or any application can be synthesised and implemented in many ways. In a multitude of applications it is desirable to use a distributed control system, with measurement and actuation tasks being performed on several nodes commuting on a shared bus. With MIMO controllers this distribution poses challenges, as it is not trivial where and how different parts of a synthesised controller should be implemented. In this thesis a mixed sensitivity approach to airspeed control of a fixed wing
UAV is explored,as well as exploring general requirements for distributing MIMO control systems.

A model of the relevant dynamics of the UAV has been created in Simulink and the model has been compared to the flight characteristics of the physical UAV. A control approach using suitable actuators, elevator and throttle, has been created and verified in the simulation environment. To implement the synthesised controller in the available hardware CAN communication and RTOS tasks have been designed and scheduled. It has been found that MIMO mixed sensitivity control can be implemented in a distributed CAN control system achieving the same functionality in the complete distributed embedded system as in a simulated controller/plant model. In practical flight tests functionality has been confirmed, achieving reliable airspeed control, both subjectively and quantitatively.

Performance in certain frequency ranges due to external factors, such as turbulence and model inaccuracies, makes direct flight comparisons to simulated response impractical. Integrating the developed airspeed control approach in a wider fully autonomous flight control scheme can be a suitable continuation.