Deteksi Batas Sungai Untuk Navigasi Perahu Menggunakan Visi Komputer

Rashida, Aqila (2025) Deteksi Batas Sungai Untuk Navigasi Perahu Menggunakan Visi Komputer. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022211114-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Deteksi batas sungai untuk navigasi perahu menggunakan visi komputer merupakan solusi alternatif untuk kondisi perairan dengan keterbatasan sinyal GPS. Pada penelitian ini, sistem dikembangkan dengan memanfaatkan model YOLOv8n-seg untuk mendeteksi batas kiri sungai, batas kanan sungai, dan ujung perahu (boat_tip) secara real-time dari citra kamera yang dipasang pada badan perahu. Proses deteksi batas sungai menggunakan visi komputer ini menjadi dasar dalam menentukan arah gerak perahu. Sistem navigasi yang dihasilkan menggunakan kontrol berbasis logika if-else sederhana, dengan perhitungan posisi relatif antara batas sungai dan posisi kapal untuk menghasilkan perintah navigasi. Model YOLOv8n-seg menunjukkan performa segmentasi yang baik dengan mAP50 sebesar 0,829, F1-score tertinggi 0,81, dan rata-rata precision di atas 0,85 serta recall di atas 0,75. Sistem diimplementasikan pada Raspberry Pi menggunakan model ONNX dengan kecepatan inferensi rata-rata 167,7 ms per frame (5,96 FPS). Dalam pengujian navigasi perahu, sistem mencatat rata-rata error posisi (0,40; 7,55) piksel pada titik boat_tip, dan (1,87; 0,46) piksel pada horizon_mid sebagai acuan jalur navigasi. Hasil ini menunjukkan bahwa deteksi batas sungai berbasis visi komputer dapat digunakan untuk memberikan arahan navigasi secara langsung kepada perahu secara efisien, meskipun masih perlu pengembangan lebih lanjut untuk mencapai sistem navigasi otonom penuh yang adaptif terhadap dinamika lingkungan sungai. =======================================================================================================================================
River edge detection for boat navigation using computer vision offers an alternative solution for waterway navigation in environments with limited GPS signals. In this study, a system was developed utilizing the YOLOv8n-seg model to detect the left riverbank, right riverbank, and the boat's front tip (boat_tip) in real-time from images captured by an onboard camera. The river edge detection using computer vision serves as the foundation for determining the boat's direction of movement. The navigation system is implemented with a simple if-else control logic, calculating the boat's trajectory based on the relative position between the detected river edges and the boat's current position. The YOLOv8n-seg model demonstrated good segmentation performance, achieving a mAP50 of 0.829, a highest F1-score of 0.81, and average precision above 0.85 and recall above 0.75. The system was deployed on a Raspberry Pi using the ONNX model format, achieving an average inference time of 167.7 ms per frame (5.96 FPS). In the boat navigation test, the system recorded an average positional error of (0.40; 7.55) pixels for the boat_tip, and (1.87; 0.46) pixels for the horizon_mid as the reference navigation point. These results indicate that river edge detection based on computer vision can efficiently provide navigation guidance to the boat in real-time. However, further development is needed to achieve a fully autonomous navigation system that adapts to dynamic river environments.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Autonomous Surface Vehicle, YOLOv8, Segementasi Citra, Visi Komputer, Image Segmentation, Computer Vision
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Electrical Technology > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Aqila Rashida
Date Deposited:	24 Jul 2025 06:38
Last Modified:	24 Jul 2025 06:38
URI:	http://repository.its.ac.id/id/eprint/121081

Actions (login required)

View Item