Please use this identifier to cite or link to this item:
Title: Structure priors aided visual-inertial navigation in building inspection tasks with auxiliary line features
Authors: Lyu, Yang
Yuan, Shenghai
Xie, Lihua
Keywords: Engineering::Electrical and electronic engineering
Issue Date: 2022
Source: Lyu, Y., Yuan, S. & Xie, L. (2022). Structure priors aided visual-inertial navigation in building inspection tasks with auxiliary line features. IEEE Transactions On Aerospace and Electronic Systems, 58(4), 3037-3048.
Journal: IEEE Transactions on Aerospace and Electronic Systems
Abstract: The article proposes a visual-inertial navigation method to support the autonomous operation of a quadrotor UAV during the building faade inspection tasks, where state-of-the-art vision-based localization methods may fail due to texture point feature insufficiency. Considering the appearance characteristics of the building faades, we additionally fuse line features and their corresponding structure prior information to the sliding-window-based estimator to improve localization reliability and accuracy. The contribution of the proposed method lies mainly in two aspects. First, we develop an informative feature selection mechanism according to the faade patterns and the inspection trajectory patterns to make a balance between localization accuracy and computation overheads. Second, we further utilize structure prior information, which is defined as line-to-line and point-to-line relationships, as another source of high-fidelity measurement to restrain the localization drifts. The proposed method is tested not only on public datasets, but also on an actual flight data package recorded in a building inspection task. Experimental results show that the proposed method can serve as a practical tool for navigating a robot in a building inspection task, with improved localization reliability and accuracy.
ISSN: 0018-9251
DOI: 10.1109/TAES.2022.3142663
Rights: © 2022 IEEE. All rights reserved.
Fulltext Permission: none
Fulltext Availability: No Fulltext
Appears in Collections:EEE Journal Articles

Page view(s)

Updated on Jan 29, 2023

Google ScholarTM




Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.