An Off-board Model Predictive Control With Obstacle Avoidance For Unmanned Aerial Vehicles

The Unmanned Aerial Vehicle technology, which was supported for the development of military technology in the early periods, has become one of the most challenging engineering research topics with the recent development of microcomputer, sensor and battery technology. The UAV field of application has been expanding relative to the increase in the capability and the feasibility of the UAV technology itself. For that reason, it is aimed to design the off-board linear model predictive control which leads the vehicle to have improved obstacle avoidance capability through the UAV path. According to goal of the project, the development of the UAV is followed from the beginning to the current stage in order to understand the role in the challenging engineering problems. The UAV concept, components and the control theory are comprehensively covered and discussed during this thesis report. The theory of the linear model predictive control, which is based on the predictions of the future steps by considering the mathematical model of the plant, is examined and combined with the Object Oriented Programming skills based on the C++ language in order to develop and improve the control of the vehicle which is selected as quadcopter for this research. Moreover, Robotic Operating System environment is used to manage the communication for the control purpose. A mathematical modelling approaches are evaluated to have proper representation of the system in order to increase the accuracy and the control. On the other hand, the obstacle avoidance capability is derived thanks to the adopted mixed linear integer programming technique which is employed in the our ROS-based implementation. For that implementation, the convex optimization procedure is operated with the selected solver which is developed in Python language thanks to the compatibility benefit of the ROS framework. Thus, the implemented approach is tested with the different target points which are selected in order to test capability limits and the results are discussed in the sense of the feasibility and performance in the last chapter.