Deep learning for vision-based micro aerial vehicle autonomous landing

Leijian Yu, Cai Luo, Xingrui Yu, Xiangyuan Jiang, Erfu Yang, Chunbo Luo, Peng Ren

Research output: Contribution to journalArticle

  • 1 Citations

Abstract

Vision-based techniques are widely used in micro aerial vehicle autonomous landing systems. Existing vision-based autonomous landing schemes tend to detect specific landing landmarks by identifying their straightforward visual features such as shapes and colors. Though efficient to compute, these schemes only apply to landmarks with limited variability and require strict environmental conditions such as consistent lighting. To overcome these limitations, we propose an end-to-end landmark detection system based on a deep convolutional neural network, which not only easily scales up to a larger number of various landmarks but also exhibit robustness to different lighting conditions. Furthermore, we propose a separative implementation strategy which conducts convolutional neural network training and detection on different hardware platforms separately, i.e. a graphics processing unit work station and a micro aerial vehicle on-board system, subject to their specific implementation requirements. To evaluate the performance of our framework, we test it on synthesized scenarios and real-world videos captured by a quadrotor on-board camera. Experimental results validate that the proposed vision-based autonomous landing system is robust to landmark variability in different backgrounds and lighting situations.

LanguageEnglish
Pages171-185
Number of pages15
JournalInternational Journal of Micro Air Vehicles
Volume10
Issue number2
Early online date16 May 2018
DOIs
StatePublished - 30 Jun 2018

Fingerprint

Landing
Lighting
Aircraft landing systems
Neural networks
Cameras
Antennas
Color
Hardware
Drones
Deep learning

Keywords

  • convolutional neural networks
  • Micro aerial vehicle
  • vision-based autonomous landing

Cite this

Yu, Leijian ; Luo, Cai ; Yu, Xingrui ; Jiang, Xiangyuan ; Yang, Erfu ; Luo, Chunbo ; Ren, Peng. / Deep learning for vision-based micro aerial vehicle autonomous landing. In: International Journal of Micro Air Vehicles. 2018 ; Vol. 10, No. 2. pp. 171-185
@article{6db8306b35e840d58712a0a653655652,
title = "Deep learning for vision-based micro aerial vehicle autonomous landing",
abstract = "Vision-based techniques are widely used in micro aerial vehicle autonomous landing systems. Existing vision-based autonomous landing schemes tend to detect specific landing landmarks by identifying their straightforward visual features such as shapes and colors. Though efficient to compute, these schemes only apply to landmarks with limited variability and require strict environmental conditions such as consistent lighting. To overcome these limitations, we propose an end-to-end landmark detection system based on a deep convolutional neural network, which not only easily scales up to a larger number of various landmarks but also exhibit robustness to different lighting conditions. Furthermore, we propose a separative implementation strategy which conducts convolutional neural network training and detection on different hardware platforms separately, i.e. a graphics processing unit work station and a micro aerial vehicle on-board system, subject to their specific implementation requirements. To evaluate the performance of our framework, we test it on synthesized scenarios and real-world videos captured by a quadrotor on-board camera. Experimental results validate that the proposed vision-based autonomous landing system is robust to landmark variability in different backgrounds and lighting situations.",
keywords = "convolutional neural networks, Micro aerial vehicle, vision-based autonomous landing",
author = "Leijian Yu and Cai Luo and Xingrui Yu and Xiangyuan Jiang and Erfu Yang and Chunbo Luo and Peng Ren",
year = "2018",
month = "6",
day = "30",
doi = "10.1177/1756829318757470",
language = "English",
volume = "10",
pages = "171--185",
journal = "International Journal of Micro Air Vehicles",
issn = "1756-8293",
number = "2",

}

Deep learning for vision-based micro aerial vehicle autonomous landing. / Yu, Leijian; Luo, Cai; Yu, Xingrui; Jiang, Xiangyuan; Yang, Erfu; Luo, Chunbo; Ren, Peng.

In: International Journal of Micro Air Vehicles, Vol. 10, No. 2, 30.06.2018, p. 171-185.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Deep learning for vision-based micro aerial vehicle autonomous landing

AU - Yu,Leijian

AU - Luo,Cai

AU - Yu,Xingrui

AU - Jiang,Xiangyuan

AU - Yang,Erfu

AU - Luo,Chunbo

AU - Ren,Peng

PY - 2018/6/30

Y1 - 2018/6/30

N2 - Vision-based techniques are widely used in micro aerial vehicle autonomous landing systems. Existing vision-based autonomous landing schemes tend to detect specific landing landmarks by identifying their straightforward visual features such as shapes and colors. Though efficient to compute, these schemes only apply to landmarks with limited variability and require strict environmental conditions such as consistent lighting. To overcome these limitations, we propose an end-to-end landmark detection system based on a deep convolutional neural network, which not only easily scales up to a larger number of various landmarks but also exhibit robustness to different lighting conditions. Furthermore, we propose a separative implementation strategy which conducts convolutional neural network training and detection on different hardware platforms separately, i.e. a graphics processing unit work station and a micro aerial vehicle on-board system, subject to their specific implementation requirements. To evaluate the performance of our framework, we test it on synthesized scenarios and real-world videos captured by a quadrotor on-board camera. Experimental results validate that the proposed vision-based autonomous landing system is robust to landmark variability in different backgrounds and lighting situations.

AB - Vision-based techniques are widely used in micro aerial vehicle autonomous landing systems. Existing vision-based autonomous landing schemes tend to detect specific landing landmarks by identifying their straightforward visual features such as shapes and colors. Though efficient to compute, these schemes only apply to landmarks with limited variability and require strict environmental conditions such as consistent lighting. To overcome these limitations, we propose an end-to-end landmark detection system based on a deep convolutional neural network, which not only easily scales up to a larger number of various landmarks but also exhibit robustness to different lighting conditions. Furthermore, we propose a separative implementation strategy which conducts convolutional neural network training and detection on different hardware platforms separately, i.e. a graphics processing unit work station and a micro aerial vehicle on-board system, subject to their specific implementation requirements. To evaluate the performance of our framework, we test it on synthesized scenarios and real-world videos captured by a quadrotor on-board camera. Experimental results validate that the proposed vision-based autonomous landing system is robust to landmark variability in different backgrounds and lighting situations.

KW - convolutional neural networks

KW - Micro aerial vehicle

KW - vision-based autonomous landing

UR - http://www.scopus.com/inward/record.url?scp=85047067219&partnerID=8YFLogxK

UR - http://journals.sagepub.com/home/mav

U2 - 10.1177/1756829318757470

DO - 10.1177/1756829318757470

M3 - Article

VL - 10

SP - 171

EP - 185

JO - International Journal of Micro Air Vehicles

T2 - International Journal of Micro Air Vehicles

JF - International Journal of Micro Air Vehicles

SN - 1756-8293

IS - 2

ER -