원문정보
초록
영어
This study was intended to find a cheap and effective method to the remote management of systems and utilities on board an unmanned vehicle running on a linux-based system as well as the determination of an effective confidence factor setting for the detection of visual markers in an augmented reality environment of ARToolKit. Opensource tools were selected due to their easy accessibility and its feature of being freely customizable that can be tweaked for our specific needs. The QGroundcontrol was chosen as the ground control system, the MAVLink as the communications protocol and the ARToolKit as the needed utility for the vision-based navigation of the system. The resulting data association problem when doing the confidence factor determination was addressed using the nearest neighbor approach. The scenario used in the test was that an operator sends a signal command using the QGroundcontrol to the unmanned vehicle through the MAVLink protocol to activate or deactivate the ARToolKit pattern tracking on-board the linux-based unmanned vehicle. The resulting tests and experiments showed that the proposed method using opensource tools performed satisfactorily as desired.
목차
1. Introduction
2. Tools used
2.1. Marker-based vision system
2.2. Ground control system for robot command
2.2. MAVLink
3. Remote Activation
4. Offline-/ online- processing and nearest neighbor method for best confidence factor setting
5. Conclusion
Acknowledgements
References