Putranto, Hamid Yusuf (2022) Identifikasi Area Pendaratan Yang Aman Pada Autonomous Delivery Uav Dengan Metode Deep Neural Network. Masters thesis, Institut Teknologi Sepuluh Nopember.
|
Text
6022201031-Master Thesis.pdf Restricted to Repository staff only Download (2MB) |
Abstract
Pengenalan area pendaratan secara otomatis membantu UAV sehingga dapat mendarat dengan tepat dan aman. Fungsi ini dapat dilakukan dengan menggunakan kamera untuk mendeteksi sebuah tanda khusus, seperti QR code, April Tag, susunan bentuk geometris, dll. Namun tidak mampu mendeteksi secara natural lingkungan di sekitarnya. Kemajuan deep learning saat ini berpotensi dapat menggantikan penanda khusus tersebut serta lebih mampu mengenali objek lain di sekitar area pendaratan lebih natural. Apalagi UAV beroperasi di daerah perkotaan/pemukiman yang rawan dengan kondisi yang tidak menentu. Sehingga, penelitian ini mengusulkan metode semantic segmentation untuk menyegmentasi objek di area pendaratan UAV dan mengidentifikasi area pendaratan yang aman. Deteksi contour digunakan untuk mengekstrak hasil segmentasi untuk mendapatkan poligon terbesar dalam contour dan poin koordinat target pendaratan. Proses prediksi setiap gambar membutuhkan waktu 0.52 detik . Metode keseluruhan ini berhasil dijalankan menggunakan prosesor i3 dengan kecepatan pemrosesan rata-rata 4 frames per second. Hasil penelitian menyatakan bahwa model memiliki kemampuan segmentasi yang baik untuk mengidentifikasi target area pendaratan UAV.
==============================================================================================================================
Identification the landing area automatically helps the UAV so that it can land correctly and safely. The UAV can do this by using the camera to detect a special sign, such as QR code, April Tag, geometric shape arrangement, etc. However, it is unable to detect the surrounding environment naturally. Currently, advances in deep learning have the potential to replace these special markers and be able to recognize other objects around the landing area more naturally. Moreover, UAVs operate in urban/settlement areas prone to uncertain conditions. Thus, this study proposes a semantic segmentation method to segment objects in the UAV (Quadcopter) landing area and identify safe landing areas. Contour detection is used to extract the segmentation results to get the largest polygon in the contour and coordinate points of the landing target. The process of predicting each image takes 0.52 seconds. This overall method was successfully executed using an i3 processor with an average processing speed of 4 frames per second. The results showed that the model has a good segmentation ability to identify the target UAV landing area.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTE 621.384 8 Put i-1 2022 |
| Uncontrolled Keywords: | identifikasi area pendaratan, semantic segmentation, Unet, UAV. landing area identification, semantic segmentation, UAV, Unet. |
| 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: | Mr. Marsudiyana - |
| Date Deposited: | 06 Jul 2026 04:27 |
| Last Modified: | 06 Jul 2026 04:27 |
| URI: | http://repository.its.ac.id/id/eprint/134339 |
Actions (login required)
![]() |
View Item |
