(For USM Staff/Student Only)

EngLib USM > Ω School of Aerospace Engineering >

Obstacle avoidance using convolutional neural network for drone navigation in oil palm plantation

Obstacle avoidance using convolutional neural network for drone navigation in oil palm plantation / Lee Hui Yin
Di Malaysia, pertanaman kelapa sawit merupakan salah satu sektor penting yang menyumbang kepada ekonomi negara. Kebelakangan ini, dron digunakan secara meluas dalam pertanian ketepatan. Walau bagaimanapun, antara cabaran untuk misi penerbangan di ketinggian rendah adalah keupayaan untuk mengelakkan perlanggaran daripada halangan untuk mengelakkan kemalangan drone. Kebanyakan sastera terdahulu menunjukkan sistem halangan pengelakan dengan sensor aktif yang biasanya tidak digunakan dalam kenderaan udara kecil disebabkan oleh kekangan kos, keberatan dan penggunaan kuasa. Dalam kajian ini, kami membentangkan satu sistem baru yang membolehkan navigasi autonomi sebuah dron kecil di ladang kelapa sawit dengan menggunakan kamera monokular sahaja. Sistem ini dibahagikan kepada dua peringkat utama: halangan pengesanan berasaskan penglihatan dan kawalan gerakan berdasarkan hasil dari peringkat pertama. Oleh sebab penglihatan monokular tidak memberikan maklumat kedalaman, antara satu teknik pembelajaran mesin, Faster R-CNN dilatih dan disesuaikan untuk pengesanan batang pokok. Selanjutnya, ketinggian kotak perbatasan yang diramalkan menganggarkan jarak halangan tersebut dari dron. Model pengesanan dinilai berdasarkan purata ketepatan dengan imej yang tidak termasuk dalam kumpulan latihan sebelum ini. Dalam sistem ini, drone diprogramkan untuk bergerak ke depan sehingga model pengesanan mengesan sebarang halangan frontal yang berhampiran. Seterusnya, arah pergerakan elakan ditakrifkan dengan mengarahkan sudut yaw berdasarkan koordinat-x yang menunjukkan arah laluan optimum yang mempunyai ruang bebas daripada halangan yang paling lebar. Kami menunjukkan prestasi sistem ini dengan melakukan ujian penerbangan dalam persekitaran ladang kelapa sawit sebenar di dua lokasi yang berbeza. Antara satu lokasi ialah lokasi yang baru. Keputusan tersebut menunjukkan bahawa kaedah yang dicadangkan itu adalah calon yang tepat dan kuat untuk navigasi autonomi dron berpandukan penglihatan di sebuah ladang kelapa sawit. _______________________________________________________________________________________________________ In Malaysia, oil palm plantation is one of the vital sectors that contribute to the country economy. In recent years, drones are widely applied in the precision agriculture due to their flexibility and capability. However, one of the challenges in a low-altitude flight mission is the ability to avoid the obstacles in order to prevent the drone crashes. Most of the previous literature demonstrated the obstacle avoidance systems with active sensors which are not applicable on small aerial vehicles due to the cost, weight and power consumption constraints. In this research, we present a novel system that enables the autonomous navigation of a small drone in the oil palm plantation using a single camera only. The system is divided into two main stages: vision-based obstacle detection, in which the obstacles in the input images are detected, and motion control, in which the avoidance decisions are taken based on the results from the first stage. As the monocular vision does not provide depth information, a machine learning model, Faster R-CNN, was trained and adapted for the tree trunk detection. Subsequently, the heights of the predicted bounding boxes were used to indicate their estimated distances from the drone. The detection model performance was validated on the testing images in term of the average precision. In the system, the drone is programmed to move forward until the detection model detects any closed frontal obstacle. Next, the avoidance motion direction is defined by commanding a yawing angle which is corresponded to the x-coordinate in the image that indicated the optimum path direction with the widest obstacle-free space. We demonstrated the performance of the system by carrying out flight tests in the real oil palm plantation environment in two different locations, where one of them is a new place. The results showed that the proposed method was accurate and robust for the drone vision-based autonomous navigation in the oil palm plantation.
Contributor(s):
Lee Hui Yin - Author
Primary Item Type:
Final Year Project
Identifiers:
Accession Number : 875008167
Language:
English
Subject Keywords:
drones; agriculture; low-altitude
First presented to the public:
6/1/2019
Original Publication Date:
7/9/2019
Previously Published By:
Universiti Sains Malaysia
Place Of Publication:
School of Aerospace Engineering
Citation:
Extents:
Number of Pages - 98
License Grantor / Date Granted:
  / ( View License )
Date Deposited
2019-07-09 15:55:27.434
Submitter:
Mohd Jasnizam Mohd Salleh

All Versions

Thumbnail Name Version Created Date
Obstacle avoidance using convolutional neural network for drone navigation in oil palm plantation1 2019-07-09 15:55:27.434