University of Twente Student Theses

Login

3D reconstruction improvement by path planning towards physical interaction with a UAV

Radl, P.D. (2019) 3D reconstruction improvement by path planning towards physical interaction with a UAV.

[img] PDF
3MB
Abstract:Physical interaction with objects by unmanned aerial vehicles (UAVs) is an emerging fi eld in current research. To interact with objects accurately, without damaging them or the UAV itself, exact representations of objects are desirable. Often predefi ned three-dimensional( 3D) models are provided to meet this constraint. To not have to rely on predefi ned 3D models, it is necessary to gain them by the UAV “on-the-fl y”. A common method to reconstruct online objects or environments is to attach sensors to UAVs and use the output as an input for simultaneous localization and mapping (SLAM) algorithms. In this research it has been investigated if the thereby created 3D models can be improved by multiple coverage. At fi rst it was necessary to determine an appropriate visual SLAM algorithm, which works reli- able with a stereo camera in a simulation. Based on this different coverage path planning (CPP) methods were evaluated if multiple coverage of areas of the object of interest can improve a resulting 3D model. Two were thereby designed as online path planning algorithms to au- tonomously determine fl ight directions. As SLAM algorithm ORB-SLAM2 by Mur-Artal and Tardos (2017) has been chosen and is shown to be suitable. Furthermore, results indicating that multiple coverage improves 3D models, represented as point cloud output of ORB-SLAM2, are provided. Moreover, potential for online path planning algorithms to improve fl ight time could be found.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:53 electrotechnology, 54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:http://purl.utwente.nl/essays/78023
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page