Sistem LiDAR-SLAM dengan Optimasi Graf Pose untuk Pemetaan 3D Akurat Menggunakan Quadcopter

Hutajulu, Aziel Godwin (2025) Sistem LiDAR-SLAM dengan Optimasi Graf Pose untuk Pemetaan 3D Akurat Menggunakan Quadcopter. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022211196-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Pemetaan tiga dimensi (3D) yang akurat dan lokalisasi mandiri di lingkungan urban tanpa sinyal GPS merupakan tantangan krusial bagi kendaraan otonom seperti quadcopter. Untuk menjawab tantangan ini, penelitian ini mengembangkan sistem LiDAR-SLAM berbasis quadcopter untuk membangun peta 3D dan memperkirakan posisi UAV secara simultan. Guna meminimalkan akumulasi drift, diterapkan metode Pose Graph Optimization (PGO) setelah proses loop closure, sementara kontrol PID bertingkat memastikan pergerakan platform yang stabil untuk akuisisi data. Sistem diuji melalui simulasi di lingkungan US City Block dalam berbagai skenario, termasuk perbedaan kompleksitas lintasan dan adanya gangguan eksternal. Hasil penelitian menunjukkan bahwa PGO secara konsisten meningkatkan akurasi dan konsistensi geometris peta, yang divalidasi dengan penurunan Root Mean Square Error (RMSE) trajektori pada setiap skenario, dengan rentang perbaikan antara 0.4 hingga 1.7. Lebih lanjut, sistem LiDAR-SLAM terbukti robust, tetap berfungsi optimal pada kondisi minim cahaya, dimana Stereo SLAM mengalami kegagalan, dan mampu mempertahankan integritas peta saat dihadapkan pada simulasi gangguan angin. Penelitian ini mendemonstrasikan bahwa kerangka kerja yang diusulkan merupakan solusi yang andal dan robust untuk pemetaan 3D akurat di lingkungan tanpa GNSS.
====================================================================================================================================
Accurate three-dimensional (3D) mapping and self-localization in urban environments without GPS signals represent a crucial challenge for autonomous vehicles like quadcopters. To address this challenge, this study develops a quadcopter-based LiDAR-SLAM system to simultaneously construct a 3D map and estimate the UAV's position. To minimize cumulative drift, a Pose Graph Optimization (PGO) method is applied after loop closure, while a nested-loop PID control ensures a stable platform movement for data acquisition. The system was tested through simulations in a US City Block environment under various scenarios, including different trajectory complexities and external disturbances. The results show that PGO consistently improves map accuracy and geometric consistency, validated by a reduction in the trajectory Root Mean Square Error (RMSE) in every scenario, with improvements ranging from 0.4 to 1.7. Furthermore, the LiDAR-SLAM system proved to be robust, remaining fully functional in low light conditions, where Stereo SLAM failed, and maintaining map integrity when subjected to simulated wind disturbances. This study demonstrates that the proposed framework is a reliable and robust solution for accurate 3D mapping in GNSS-denied environments.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Quadcopter, LiDAR-SLAM, Pose Graph Optimization, Kontrol PID, Quadcopter, LiDAR-SLAM, Pose Graph Optimization, PID Control
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3070 Automatic control T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL776 .N67 Quadrotor helicopters--Automatic control
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Aziel Godwin Hutajulu
Date Deposited:	25 Jul 2025 06:50
Last Modified:	25 Jul 2025 06:50
URI:	http://repository.its.ac.id/id/eprint/121510

Actions (login required)

View Item