Developing An Optical and Patern Recognition Based Precission Landing System For Rotary wing UAVS


Thesis Type: Postgraduate

Institution Of The Thesis: Marmara University, Faculty Of Technology, Mechatronic Engineering, Turkey

Thesis Language: Turkish

Student: Kübra PEHLİVAN

Principal Supervisor (For Co-Supervisor Theses): Hasan Erdal

Co-Supervisor: Barış Doğan

Abstract:

As the use of UAVs has become widespread, there has created need for autonomous and precision landing on a specific destination on water, indoor and outdoor areas on land. Today, approaches to solve the precision landing depends on Global Navigation Satellite System (GNSS), Inertial Navigation System (INS) or Real-Time Kinematics (RTK) technologies etc. Most of these technologies are high cost and in INS and GNSS systems the magnitude of position error in the landing on target can be large. The use of measurement systems based on image data in the control and landing applications of UAVs is becoming widespread with the development of computer vision and signal processing based technologies.

In this thesis, development of optical based autonomous precision landing system is aimed for rotary wing UAVs which are capable of vertical land / take off. In this way, the manufacture of quadrotor which has open source flight control unit (FCU) and four rotary propeller, conducting  detection tests of light at different wavelengths in different conditions, using image processing algorithms with camera and position / speed based PID position control  are purposed.

In many similar studies about precision landing, Red-Green-Blue (RGB) or grayscale color space-based methods are used in the daylight band. The landing on target position is determined by methods of recognition the colors or shape patterns on the surface of target in RGB color space based applications. In grayscale color space-based applications, markers such as ARUCO, AprilTag etc. which have prebuild libraries for programming languages are used to determine the target landing position. Following this, applications that is studied on daylight band have problems such as change of color values, target cannot be recognized or seen as a result of changing light intensity at different times of day.

In this study, in order to prevent the problems encountered in daylight band-based applications, Beacon design will be did which works in IR band and has active light  source. In order to detect the designed active pointer, Image processing algorithms will be developed by using camera which is only work in IR band. Precision landing system based on image processing will be developed with Python language on one of the popular mini-computer platforms used today. The precision landing system consisting of mini-computer, IR camera, IR pointer and control application will be commercialized for many rotary wing UAVs. In this respect, the study can also be transformed into TUBITAK 1512 project applications which may contribute economically.

With the design studies to be done on the pattern model of the IR marke, the detection distance of the marker will be increased in all light conditions. Furthermore with the algorithm that will be developed in the control application, the landing accuracy of the UAV will be improved to a better level than the existing applications. All studies that are conducted in this direction will constitute the original aspect of this thesis proposal.