Implementasi Yolo V8 pada Prototipe Autonomous Underwater Robot Berbasis Raspberry Pi 5 Guna Menanggulangi Pencemaran Sampah Plastik di Daerah Perairan

Jonathan Widodo Wiji Saputra (Universitas Negeri Malang, Indonesia)
Muhammad Allam Naufal (Universitas Negeri Malang, Indonesia)
Oditya Andalas Putra (Universitas Negeri Malang, Indonesia)

jurnal syntax admiration
Vol. 5 No. 11 (2024): Jurnal Syntax Admiration

Plastic waste pollution in aquatic environments threatens marine ecosystems and human health. This study developed a prototype Autonomous Underwater Robot (AUR) based on Raspberry Pi 5, equipped with the YOLO V8 algorithm for real-time detection and classification of plastic waste. Additionally, the AUR uses a PID control system that enables movement stabilization and accurate navigation towards the target. Testing was conducted to evaluate object detection accuracy using mAP and IoU metrics, as well as the PID control performance in maintaining orientation stability. Results indicate that YOLO V8 on the AUR can detect and classify objects with an mAP of 0.6772 at IoU 0.5 and a detection accuracy of 89.6%. The PID control system, with an optimal parameter setting (Kp:Ki:Kd = 125:12:8), achieved an object center accuracy of 96.4%, orientation stability of 90.8%, and a completion time of 2 seconds, demonstrating the efficiency of the AUR in addressing plastic waste in aquatic environments effectively and sustainably.