UAS Navigation with SqueezePoseNet—Accuracy Boosting for Pose Regression by Data Augmentation

Mueller, Markus; Jutzi, Boris

doi:10.3390/drones2010007

Cited by 8 publications

(4 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of these is indoor navigation or operation in urban canyon environments where there are many high-rise buildings around the receiver. When there are obstacles blocking the line-of-sight (LOS) signal between the receiver and the user, GNSS signals may have large amounts of multipath due to signals bouncing off structures, which leads to inaccurate ranging measurements, or may be significantly attenuated or unavailable [14]. The second and more troubling limitation is operation in the presence of RFI, such as spoofing or jamming.…”

Section: Gnssmentioning

confidence: 99%

Assessment of Android Network Positioning as an Alternative Source of Navigation for Drone Operations

Lee

Nedelkov

Akos

2022

Drones

View full text Add to dashboard Cite

Applications of drones have increased significantly in the past decade for both indoor and outdoor operations. In order to assist autonomous drone navigation, there are numerous sensors installed onboard the vehicles. These include Global Navigation Satellite Systems (GNSS) chipsets, inertial sensors, barometer, lidar, radar and vision sensors. The two sensors used most often by drone autopilot controllers for absolute positioning are the GNSS chipsets and barometer. Although, for most outdoor operations, these sensors provide accurate and reliable position information, their accuracy, availability, and integrity deteriorate for indoor applications and in the presence of radio frequency interference (RFI), such as GNSS spoofing and jamming. Therefore, it is possible to derive network-based locations from Wi-Fi and cellular transmission. Although there have been many theoretical studies on network positioning, limited resources are available for the expected quantitative performance of these positioning methodologies. In this paper, the authors investigate both the horizontal and vertical accuracy of the Android network location engines under rural, suburban, and urban environments. The paper determines the horizontal location accuracy to be approximately 1637 m, 38 m, and 32 m in terms of 68% circular error probable (CEP) for rural, suburban, and urban environments, respectively, and the vertical accuracy to be 1.2 m and 4.6 m in terms of 68% CEP for suburban and urban environments, respectively. In addition, the availability and latency of the location engines are explored. Furthermore, the paper assesses the accuracy of the Android network location accuracy indicator for various drone operation environments. The assessed accuracies of the network locations provide a deeper insight into their potential for drone navigation.

show abstract

Section: Gnssmentioning

confidence: 99%

Assessment of Android Network Positioning as an Alternative Source of Navigation for Drone Operations

Lee

Nedelkov

Akos

2022

Drones

View full text Add to dashboard Cite

show abstract

“…Recently, significant results have been achieved by Deep Neural Networks (DNN) in the task of pose estimation based on monocular imagery. In this sense, the use of a Convolutional Neural Network (CNN) [6] to learn and to match features, which aids in camera pose estimation, has become popular with the work of Kendall et al [12] and more recently the work of Mueller et al [30]. However, both approaches rely on prior environmental knowledge before yielding an estimation of the camera position.…”

Section: Related Workmentioning

confidence: 99%

Multi-Task Regression-Based Learning for Autonomous Unmanned Aerial Vehicle Flight Control Within Unstructured Outdoor Environments

Maciel-Pearson

Akçay

Atapour-Abarghouei

et al. 2019

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

Increased growth in the global Unmanned Aerial Vehicles (UAV) (drone) industry has expanded possibilities for fully autonomous UAV applications. A particular application which has in part motivated this research is the use of UAV in wide area search and surveillance operations in unstructured outdoor environments. The critical issue with such environments is the lack of structured features that could aid in autonomous flight, such as road lines or paths. In this paper, we propose an End-to-End Multi-Task Regression-based Learning approach capable of defining flight commands for navigation and exploration under the forest canopy, regardless of the presence of trails or additional sensors (i.e. GPS). Training and testing are performed using a software in the loop pipeline which allows for a detailed evaluation against state-of-the-art pose estimation techniques.Our extensive experiments demonstrate that our approach excels in performing dense exploration within the required search perimeter, is capable of covering wider search regions, generalises to previously unseen and unexplored environments and outperforms contemporary state-of-the-art techniques.

show abstract

“…It was shown that pose regression in areas with less training data scores worse compared to areas with a dense distribution of training samples (Mueller et al, 2017). Utilizing a photorealistic model for data augmentation showed improvements regarding estimation accuracy (Mueller and Jutzi, 2018). However, photo-realistic models are not as wide distributed or available as triangulated 3D model.…”

Section: Related Workmentioning

confidence: 99%

CNN-Based Initial Localization Improved by Data Augmentation

Mueller

Metzger

Jutzi

2018

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Image-based localization or camera re-localization is a fundamental task in computer vision and mandatory in the fields of navigation for robotics and autonomous driving or for virtual and augmented reality. Such image pose regression in 6 Degrees of Freedom (DoF) is recently solved by Convolutional Neural Networks (CNNs). However, already well-established methods based on feature matching still score higher accuracies so far. Therefore, we want to investigate how data augmentation could further improve CNN-based pose regression. Data augmentation is a valuable technique to boost performance on training based methods and wide spread in the computer vision community. Our aim in this paper is to show the benefit of data augmentation for pose regression by CNNs. For this purpose images are rendered from a 3D model of the actual test environment. This model again is generated by the original training data set, whereas no additional information nor data is required. Furthermore we introduce different training sets composed of rendered and real images. It is shown that the enhanced training of CNNs by utilizing 3D models of the environment improves the image localization accuracy. The accuracy of pose regression could be improved up to 69.37% for the position component and 61.61% for the rotation component on our investigated data set.

show abstract

UAS Navigation with SqueezePoseNet—Accuracy Boosting for Pose Regression by Data Augmentation

Cited by 8 publications

References 52 publications

Assessment of Android Network Positioning as an Alternative Source of Navigation for Drone Operations

Assessment of Android Network Positioning as an Alternative Source of Navigation for Drone Operations

Multi-Task Regression-Based Learning for Autonomous Unmanned Aerial Vehicle Flight Control Within Unstructured Outdoor Environments

CNN-Based Initial Localization Improved by Data Augmentation

Contact Info

Product

Resources

About