Design, modelling and control of a nano quadrotor withmicrocontroller based vision system for object tracking

Abstract

The quadrotor industry is developing rapidly thanks to the advancements in many technological fields like control systems, navigation systems, communication systems and so on. The improvements in different technological fields allow the quadrotors to be used in different areas like security, delivery, military, entertainment et cetera. Another very rapidly growing sector is machine vision and artificial intelligence technology. With the latest developments in vision systems, machines can see their environment and detect and recognize objects around them and with the help of artificial intelligence they can take actions accordingly just like humans. So, the combination of quadrotor and machine vision technologies has enormously big area of application. In this thesis study, a nano-quadrotor with a microcontroller-based vision system was developed for the researchers and the students who work on control and image processing fields as a compact, inexpensive and practical solution. The developed nano-quadrotor has two microcontrollers on it where one of them is dedicated for the image processing and the other one is dedicated for the control and other tasks. These microcontrollers are on a single PCB board which is used as the body of the nano-quadrotor. The arms of the quadrotor were produced by 3D printer and coreless motors were preferred in order to make the quadrotor light weight. In the first chapter of this thesis, a brief introduction about the quadrotors and the image processing was given. In the second chapter, some of the previous researches and studies that are related to this thesis were briefly mentioned. In the third chapter, the hardware design of the nano-quadrotor was presented. The electronical design and components of the main board, the design of the arms and other selected parts were explained. In the fourth chapter, the mathematical model of the quadrotor was derived. The nonlinear model was derived and shown in state-space representation. Lastly the model was linearized to use linear controller. In chapter five, the control system design of the quadrotor was presented. The designed PID controllers, sensor fusion and state estimation systems and parameter identification processes were given. In chapter six, the general idea of image processing and its usage in microcontrollers were explained. The suitable image processing methods were discussed. Finally, the proposed image processing method for object tracking and its usage in the control system were presented. In chapter seven, the conducted tests and their results for evaluating the system performance were given. The flight scenarios were explained and the recorded flight data was shown in graphs. Lastly in chapter eight, the system performance was evaluated based on the conducted test results and some improvement suggestions for future work were given.