Pengembangan Robot Quadruped-Legged Untuk Estimasi Posisi Komponen Overheat Pada Gardu Induk Berbasis Kamera Termal

Darmawan, I Wayan Agus Darmawan (2025) Pengembangan Robot Quadruped-Legged Untuk Estimasi Posisi Komponen Overheat Pada Gardu Induk Berbasis Kamera Termal. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 5024211070-Undergraduate_Thesis.pdf] Text
5024211070-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (5MB) | Request a copy

Abstract

Peningkatan populasi dan pertumbuhan industri yang pesat menyebabkan permintaan en- ergi listrik di Indonesia terus mengalami lonjakan signifikan setiap tahunnya. Kondisi ini me- nuntut penguatan infrastruktur kelistrikan, terutama pada gardu induk yang berperan penting dalam mendistribusikan dan mengendalikan aliran listrik. Salah satu permasalahan yang ser- ing terjadi di gardu induk adalah overheat pada komponen kritis seperti transformator, isolator, dan disconnector, yang dapat mengganggu stabilitas sistem dan menimbulkan risiko kerusakan serius. Penelitian ini mengembangkan sistem robot berkaki empat (quadruped-legged robot) yang dirancang untuk mendeteksi komponen gardu induk yang mengalami overheat. Sistem ini mengintegrasikan model deteksi objek berbasis YOLOv8 untuk mengenali komponen listrik, serta menganalisis suhu pada area hasil deteksi guna mengidentifikasi potensi kondisi overheat. Navigasi robot dikendalikan menggunakan kombinasi algoritma PID dan Pure Pursuit agar mampu mengikuti lintasan patroli secara otonom, sementara proses pemetaan dan lokalisasi dilakukan secara simultan menggunakan algoritma FastLIO2 untuk menghasilkan peta awan titik tiga dimensi dan estimasi posisi yang akurat. Sistem control station dikembangkan un- tuk menerima dan memvisualisasikan data posisi, status sensor, serta citra termal secara waktu nyata menggunakan protokol WebSocket dan WebRTC. Hasil pengujian menunjukkan bahwa sistem berhasil mendeteksi overheat dengan akurasi tinggi, ditandai dengan nilai mean average precision (mAP) sebesar 0,8554 dan presisi sebesar 0,9529 untuk dua kelas komponen. Pen- gujian navigasi menunjukkan galat posisi rata-rata antara 0,40 hingga 0,85 meter, bergantung pada parameter PID dan nilai lookahead. Transmisi data berjalan stabil dengan rata-rata delay 30–50 ms untuk data status dan 210–350 ms untuk video streaming dalam jarak hingga 100 me- ter. Pengujian simulasi menggunakan data rosbag juga membuktikan kemampuan sistem dalam mendeteksi overheat dan mengirimkan data ke antarmuka web secara berhasil. Hasil ini menun- jukkan bahwa sistem pemantauan bergerak berbasis robot quadruped layak diimplementasikan sebagai solusi inspeksi otonom pada gardu induk, dengan potensi besar dalam meningkatkan efisiensi dan keamanan infrastruktur kelistrikan nasional.
===================================================================================================================================
The increasing demand for electrical energy in Indonesia requires a more reliable and intel- ligent approach to maintaining substation infrastructure. Overheating in high-voltage substation components can cause equipment failure and service disruption, making real-time monitoring essential. This research develops a quadruped-legged robot system designed to detect overheat- ing components in electrical substations using a thermal camera and onboard intelligence. The system integrates a YOLOv8-based object detection model to identify electrical components and performs temperature analysis within the detected regions of interest to determine possi- ble overheat conditions. The navigation system combines PID control with the Pure Pursuit algorithm, allowing the robot to follow predefined patrol routes while recording thermal data at inspection points. Mapping is performed using the FastLIO2 SLAM algorithm, producing a 3D point cloud of the environment. However, due to the lack of visual diversity in substation structures, the localization process relies on odometry with fixed initial positioning. A control station is implemented to monitor and receive real-time data via WebSocket and WebRTC, in- cluding position, sensor status, and thermal images. The system achieved detection accuracy with a mean average precision (mAP) of 0.8554 and a precision of 0.9529 for two classes of components. Navigation tests yielded average position errors ranging from 0.40 to 0.85 meters depending on the PID parameters and lookahead values. Data transmission remained stable with average delays of 30–50 ms for telemetry and 210–350 ms for video streaming over dis- tances of up to 100 meters. Simulation testing using ROS bag data demonstrated the system’s ability to detect overheating conditions and predict component positions, with data successfully transmitted to the web-based interface. This study demonstrates the feasibility of implementing a mobile quadruped monitoring system for autonomous inspection in substation environments, offering potential for safer and more efficient thermal diagnostics.

Item Type: Thesis (Other)
Uncontrolled Keywords: Overheat, robot quadruped, kamera termal, YOLOv8, FastLIO2, thermal camera
Subjects: Q Science > QA Mathematics > QA336 Artificial Intelligence
Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
T Technology > TJ Mechanical engineering and machinery > TJ211 Robotics.
T Technology > TJ Mechanical engineering and machinery > TJ211.415 Mobile robots
T Technology > TJ Mechanical engineering and machinery > TJ213 Automatic control.
T Technology > TJ Mechanical engineering and machinery > TJ223 PID controllers
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User: I Wayan Agus Darmawan
Date Deposited: 30 Jul 2025 02:58
Last Modified: 30 Jul 2025 02:58
URI: http://repository.its.ac.id/id/eprint/123032

Actions (login required)

View Item View Item