Sistem Penghindaran Rintangan pada Unmanned Surface Vehicle Berbasis Segmentasi Permukaan Air dan Prediksi Posisi

Nugraha, Ditya Garda (2025) Sistem Penghindaran Rintangan pada Unmanned Surface Vehicle Berbasis Segmentasi Permukaan Air dan Prediksi Posisi. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022241125-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Seiring peningkatan peran Unmanned Surface Vehicle (USV) dalam berbagai aplikasi maritim, pengembangan sistem navigasi otonom yang andal dan aman untuk menghindari rintangan menjadi sebuah kebutuhan yang penting. Penelitian ini menyajikan perancangan, implementasi, dan analisis perbandingan terhadap beberapa algoritma penghindar rintangan reaktif, yang mencakup Artificial Potential Field (APF), Follow the Gap Method (FGM), serta metode Vector Field Histogram (VFH, VFH+, dan VFH*). Tujuan utama dari penelitian ini adalah untuk memvalidasi kinerja simulasi dari data real environtment dan mengevaluasi hasil kinerja dari algoritma-algoritma tersebut dalam beberapa skenario rintangan navigasi. Metodologi yang digunakan menggunakan platform simulasi modular dengan Python, di mana lingkungan simulasi direkonstruksi dari data sensor dunia nyata. Pada tahap pra-pemrosesan, model segmentasi CGNet menganalisis citra visual dari kamera stereo untuk melakukan identifikasi fitur air dan bukan air, yang hasilnya kemudian difusikan dengan data GPS untuk menghasilkan peta berbasis local-coordinate cartesian. Hasil penelitian menunjukkan algoritma penghindar rintangan VFH+ menjadi satu-satunya metode yang berhasil mengatasi skenario jebakan kompleks dibandingkan metode lainnya dengan memperhitungkan ukuran fisik robot yang mendapatkan tingkat keberhasilan sebesar 90,4%. Disimpulkan bahwa kerangka kerja simulasi ini efektif untuk validasi algoritma, dan keberhasilan navigasi tidak hanya bergantung pada kecanggihan algoritma penghindar rintangan, tetapi juga sangat ditentukan oleh akurasi dari model persepsi CGNet dengan hasil mIoU sebesar 0,8799 yang menyediakan data segmentasi yang presisi pada skenario pengujian lingkungan nyata.
=====================================================================================================================================
As the role of Unmanned Surface Vehicles (USVs) increases in various maritime applications, the development of reliable and safe autonomous navigation systems for obstacle avoidance has become an important necessity. This study presents the design, implementation, and comparative analysis of several reactive obstacle avoidance algorithms, including Artificial Potential Field (APF), Follow the Gap Method (FGM), and Vector Field Histogram methods (VFH, VFH+, and VFH*). The main objective of this research is to validate the simulation performance using real environment data and evaluate the performance results of these algorithms in several obstacle navigation scenarios. The methodology uses a modular simulation platform with Python, where the simulation environment is reconstructed from real-world sensor data. In the pre-processing stage, the CGNet segmentation model analyzes visual images from a stereo camera to identify water and non-water features, whose results are then fused with GPS data to generate a local-coordinate Cartesian-based map. The results show that the VFH+ obstacle avoidance algorithm is the only method that successfully overcomes complex trap scenarios compared to other methods by considering the physical size of the robot, achieving a success rate of 90.4%. It is concluded that this simulation framework is effective for algorithm validation, and that navigation success depends not only on the sophistication of the obstacle avoidance algorithm but is also strongly determined by the accuracy of the CGNet perception model, which achieved an mIoU of 0.8799, providing precise segmentation data in real-world test scenarios.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Artificial Potential Field, CGNet, Follow the Gap-Method, Obstacle Avoidance, Penghindar Rintangan, Unmanned Surface Vehicle, Vector Field Histogram
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ211 Robotics. T Technology > TJ Mechanical engineering and machinery > TJ211.4 Robot motion T Technology > TK Electrical engineering. Electronics Nuclear engineering T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Nugraha Ditya Garda
Date Deposited:	24 Jul 2025 08:28
Last Modified:	24 Jul 2025 08:28
URI:	http://repository.its.ac.id/id/eprint/121485

Actions (login required)

View Item