Prambudi, Muhammad Rochman Thoriq (2024) Robot Pembersih Pantai berbasis AI (Artificial Intelligence). Other thesis, Institut Teknologi Sepuluh Nopember.
Text
2042201125-Undergraduated_Thesis for D4 Program.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (6MB) | Request a copy |
Abstract
Fenomena pantai dengan kondisi kotor menyebabkan berbagai dampak buruk. Maka dari itu kondisi laut yang tercemar oleh sampah dapat menimbulkan berbagai masalah seperti mengganggu kesehatan makhluk hidup sekitar, keindahan ekosistem pantai, dan dapat menimbulkan abrasi pada bibir pantai. Dibuatlah salah satu solusi untuk mengatasi permasalahan tersebut dengan membuat sebuah robot pembersih pantai berbasis AI (Artificial Intelligence). Robot ini dilengkapi dengan metode deteksi objek berupa YOLO (You Only Look Once) dengan versi YOLOv3-Tiny untuk mendeteksi sampah yang akan dikumpulkan pada pesisir pantai. Pada robot pembersih pantai berbasis AI (Artificial Intelligence) ini menggunakan Raspberry Pi 4 sebagai mikroprosesor untuk kotrol utama dan pendeteksian objek menggunakan media webcam dari Logitech serta L298N driver motor mengendalikan motor DC untuk jalannya robot. Pada tugas akhir ini juga dilengkapi terdapat GUI yang tersedia di smartphone untuk mengontrol pergerakan serta memantau deteksi objek saat robot berjalan. Training YOLO dengan dataset sebanyak 1001 gambar sampah pada pantai yang telah dilakukan menghasilkan beberapa nilai parameter pada iterasi ke-4000 seperti mAP : 0,878, Accuracy : 0,816, Recall : 0,852, dan Precision : 0,872. Pada saat training juga menghasilkan nilai loss vs iterations sebesar 1,5278 dan validasi pembacaan deteksi objek sebesar 93%. Pengujian pembacaan deteksi objek menggunakan beberapa variasi kondisi seperti ukuran dimensi sampah yang menghasilkan confidence score sebesar 0,82 dan kondisi pencahayaan kurang menghasilkan confidence score sebesar 0,58. Pada variasi lainnya pendeteksian objek dengan ukuran dimensi besar menghasilkan confidence score sebesar 0,87 dan kondisi dimensi kecil menghasilkan confidence score sebesar 0,56. Robot dapat beroperasi dan mengumpulkan sampah sebanyak 110 sampah selama satu minggu pengoperasian.
===================================================================================================================================
The phenomenon of dirty beaches causes various adverse effects. The condition of the sea polluted by waste can lead to various problems such as disturbing the health of surrounding living creatures, ruining the beauty of beach ecosystems, and causing abrasion on the coastline. One solution to address these issues is the creation of an AI-based beach cleaning robot. This robot is equipped with an object detection method using YOLO (You Only Look Once) with the YOLOv3-Tiny version to detect waste to be collected on the beach. This AI-based beach cleaning robot uses a Raspberry Pi 4 as the main microprocessor for control, object detection using a Logitech webcam, and an L298N motor driver to control the DC motor that moves the robot. This project also includes a GUI available on a smartphone to control the movement and monitor object detection while the robot operates. Training YOLO with a dataset of 1001 images of beach waste resulted in several parameter values at iteration 4000, such as mAP: 0.878, Accuracy: 0.816, Recall: 0.852, and Precision: 0.872. During training, it also produced a loss vs iterations value of 1.5278 and object detection validation reading of 93%. Object detection reading tests using various conditions, such as waste size dimensions, resulted in a confidence score of 0.82, and low lighting conditions resulted in a confidence score of 0.58. In other variations, object detection with large size dimensions resulted in a confidence score of 0.87, and small dimensions resulted in a confidence score of 0.56. The robot can operate and collect 110 pieces of waste during one week of operation.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Artificial Intelligence, Deteksi Objek, Deep Learning, Robot Pembersih Pantai, YOLO (You Only Look Once).Artificial Intelligence, Beach Cleaning Robot, Deep Learning, YOLO (You Only Look Once), Object Detection. |
Subjects: | T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques |
Divisions: | Faculty of Vocational > Instrumentation Engineering |
Depositing User: | Muhammad Rochman Thoriq Prambudi |
Date Deposited: | 30 Jul 2024 14:43 |
Last Modified: | 30 Jul 2024 14:43 |
URI: | http://repository.its.ac.id/id/eprint/110039 |
Actions (login required)
View Item |