Sistem Pendaratan Otomatis Quadcopter Pada Platform Bergerak Menggunakan Adaptive Convolutional Neural Network Sebagai Kontroler Kecepatan

Prabowo, Doni Yudi (2023) Sistem Pendaratan Otomatis Quadcopter Pada Platform Bergerak Menggunakan Adaptive Convolutional Neural Network Sebagai Kontroler Kecepatan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111940000033-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (3MB) | Request a copy

Abstract

Quadcopter merupakan salah satu jenis UAV banyak digunakan dengan sistem yang bersifat nonlinier. Salah satu metode kontrol yang efektif dan banyak digunakan untuk plant nonlinear, khususnya quadcopter adalah Neural Network. Convolutional Neural Network (CNN) yang memiliki mekanisme yang mirip dengan Neural Network tentu juga memiliki potensi sebagai metode kontrol plant nonlinear. Meskipun CNN masih kurang umum digunakan untuk metode kontrol pada real plant, terdapat suatu cara dengan memproses error yang dijadikan sebagai suatu gambar dan tambahan metode adaptive control yaitu berupa adaptive leraning rate yang dilanjutkan dengan updating weight dengan tujuan untuk mengatasi perubahan kondisi dan error pemrosesan dari model CNN tradisional sehingga menghasilkan Adaptive Convolutional Neural Network. Setelah dilakukan penelitian metode Adaptive CNN sebagai kontroler kecepatan dalam misi pendaratan otomatis pada platform bergerak, didapatkan hasil bahwa metode Adaptive Convolutional Neural Network dapat mengurangi error posisi dari quadcopter ketika pengujian sistem pendaratan otomatis. Hal tersebut dibuktikan dengan nilai Mean Absolut Error (MAE) lintasan dari Adaptive CNN lebih kecil dibandingkan dengan kontroler CNN tradisional pada simulasi dan kontroler PID pada implementasi serta dibuktikan juga melalui posisi akhir quadcopter yang lebih dekat dari platform pendaratan pada implementasi yaitu 10,3979 cm pada sumbu X, 14,1374 cm pada sumbu Y, dan 5,0537 cm pada sumbu Z pada platform pendaratan.
==================================================================================================================================
Quadcopter is a type of UAV widely used with nonlinear systems. One of the most effective and widely used control methods for nonlinear plants, especially quadcopters, is a Neural Network. Convolutional Neural Network (CNN) which has a mechanism similar to a Neural Network certainly also has the potential as a nonlinear plant control method. Although CNN is still not commonly used for control methods in real plants, there is a way to process errors that are used as an image and additional adaptive control methods, namely in the form of adaptive learning rates followed by updating weights with the aim of overcoming condition changes and processing errors from traditional CNN models so as to produce an Adaptive Convolutional Neural Network. After conducting research on the Adaptive CNN method as a speed controller in an automatic landing mission on a moving platform, it was found that the Adaptive Convolutional Neural Network method can reduce the position error of the quadcopter when testing the automatic landing system. This is evidenced by the value of the Mean Absolute Error (MAE) trajectory of the Adaptive CNN which is smaller than the traditional CNN controller in the simulation and the PID controller in the implementation and also proven by the final position of the quadcopter which is closer to the landing platform in the implementation, namely 10.3979 cm on the X axis, 14.1374 cm on the Y axis, and 5.0537 cm on the Z axis on the landing platform.

Item Type:	Thesis (Other)
Additional Information:	RSE 629.133 Pra a-1 2023
Uncontrolled Keywords:	Adaptive Learning Rate, Automatic Landing, Convolutional Neural Network, Pendaratan Otomatis, Quadcopter
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ217 Adaptive control systems T Technology > TJ Mechanical engineering and machinery > TJ223 PID controllers T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL776 .N67 Quadrotor helicopters--Automatic control U Military Science > UG1242 Drone aircraft--Control systems. (unmanned vehicle)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Doni Yudi Prabowo
Date Deposited:	29 Jul 2023 12:35
Last Modified:	06 Oct 2023 04:13
URI:	http://repository.its.ac.id/id/eprint/99949

Actions (login required)

View Item