Model Predictive Control (MPC) has been well established and widely used in the
process control industry since years. However, due to dependability of its success on availability
of high computational power to handle burden of online repetitive calculations, and
existence of a precise mathematical model of the controlled plant, it has found less application
in other areas of systems and control, specifically speaking when it comes to fast
dynamics control systems featuring a highly elaborate plant.
Preceded by previous successful efforts made in the application of MPC to other
areas of systems and control rather than process control, this thesis initiates employment
of MPC in the unmanned aerial systems industry. To this end, the system of the quadrotor
UAV testbed in the Networked Autonomous Vehicles Laboratory of Concordia University
is chosen. A three dimensional autopilot control system within the framework of MPC
is developed and tested through numerous flight experiments. The overall performance
of the quadrotor helicopter is evaluated under autonomous fight for three flight scenarios
of trajectory tracking, payload drop, robustness to voltage/current drop, and fault-tolerant
control in the presence of faults induced by reduced actuator effectiveness. This has been
achieved by the proper use of a model reduction technique as well as a fast optimization
algorithm to address the issues with high computation, and incorporation of the integral
action control in the MPC formulation to meet the offset-free tracking requirement. Both
simulation and experimental results are presented to demonstrate success of the design.