Navigasi UAV Bebas Tabrakan Di Lingkungan Bawah Kanopi Menggunakan Fusi Sensor Vision-Tof Offboard

Purnomo, Widi Saputro (2026) Navigasi UAV Bebas Tabrakan Di Lingkungan Bawah Kanopi Menggunakan Fusi Sensor Vision-Tof Offboard. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022241097_Widi Saputro_Repositori.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Navigasi Unmanned Aerial Vehicle (UAV) di lingkungan bawah kanopi menghadapi tantangan spasial yang signifikan akibat ketiadaan sinyal GPS serta keterbatasan spesifikasi perangkat keras wahana berskala mikro dalam mengeksekusi algoritma pemrosesan citra secara mandiri (onboard). Untuk mengatasi kendala tersebut, penelitian ini bertujuan mengembangkan sistem navigasi UAV bebas tabrakan (collision-free) berbasis arsitektur komputasi offboard menggunakan skema fusi sensor. Sistem bekerja dengan mendelegasikan pemrosesan sensor Vision (kamera monokuler) ke stasiun kendali darat menggunakan algoritma YOLO untuk melokalisasi koordinat celah kanopi, yang kemudian difusikan secara asinkron dengan data jarak vertikal absolut dari sensor Time of Flight (ToF). Output fusi tersebut dikonversi menjadi standar metrik melalui kalibrasi empiris untuk diumpankan sebagai nilai galat spasial pada sistem kendali Proporsional. Berdasarkan ekstraksi data telemetri, arsitektur offboard terbukti efisien dengan kecepatan inferensi stabil pada 49 Frames Per Second (FPS) dan tingkat retensi pelacakan target mencapai 100%. Lebih lanjut, skema fusi sensor memandu wahana secara akurat (di mana 150 piksel ekuivalen dengan 32 cm) dan sukses mengeksekusi manuver pelolosan dari himpitan dahan proksimal 64 cm menuju koridor udara bebas dengan ruang atap (roof clearance) yang aman pada jarak 277 cm. Sinergi arsitektur ini memastikan UAV mampu menyelesaikan misi navigasi di lingkungan kanopi secara otonom dan terhindar dari benturan fisik.
===================================================================================================================================
Unmanned Aerial Vehicle (UAV) navigation in under-canopy environments faces significant spatial challenges due to the absence of GPS signals and the hardware limitations of micro-scale vehicles in executing image processing algorithms onboard. To overcome these constraints, this study aims to develop a collision-free UAV navigation system based on an offboard computing architecture using a sensor fusion scheme. The system operates by delegating the processing of the Vision sensor (monocular camera) to the ground control station using the YOLO algorithm to localize the coordinates of canopy gaps, which are then asynchronously fused with absolute vertical distance data from the Time of Flight (ToF) sensor. The fusion output is converted into a metric standard through empirical calibration to be fed as a spatial error value into the Proportional control system. Based on the extraction of telemetry data, the offboard architecture proved efficient with a stable inference speed of 49 Frames Per Second (FPS) and a target tracking retention rate reaching 100%. Furthermore, the sensor fusion scheme guided the vehicle accurately (where 150 pixels are equivalent to 32 cm) and successfully executed an escape maneuver from a proximal branch confinement of 64 cm to a free air corridor with a safe roof clearance at a distance of 277 cm. The synergy of this architecture ensures that the UAV is capable of completing navigation missions in canopy environments autonomously while avoiding physical collisions.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Fusi Sensor, Navigasi UAV, Offboard, Time of Flight, Vision. ======================================================================================================================== Sensor Fusion, UAV Navigation, Offboard, Time of Flight, Vision.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Widi Saputro Purnomo
Date Deposited:	31 Jul 2026 01:02
Last Modified:	31 Jul 2026 01:02
URI:	http://repository.its.ac.id/id/eprint/140150

Actions (login required)

View Item