DocumentCode
2228483
Title
Demonstration of a vision-based dead-reckoning system for navigation of an underwater vehicle
Author
Huster, Andreas ; Fleischer, Stephen D. ; Rock, Stephen M.
Author_Institution
Aerosp. Robotics Lab., Stanford Univ., CA, USA
Volume
1
fYear
1998
fDate
28 Sep-1 Oct 1998
Firstpage
326
Abstract
Describes a dead-reckoning navigation system for hover-capable underwater vehicles operating close to the ocean floor. Navigation is presented as an extension of underwater station-keeping and mosaicing. It combines real-time vision-processing to build a mosaic of the area of interest, an image-based user interface to specify desired vehicle locations, and vision-based dead-reckoning to compute the robot´s position in the mosaic. This system provides a high-level interface between the vehicle and the pilot, who specifies the goal (e.g., go to and hover over this feature) instead of the commands to execute the task (e.g., rotate to the left, go forward, stop). Thus, it is an enabling technology for autonomous underwater vehicles-for which commanding actuators directly is not feasible-and a useful high-level interface for remotely operated vehicles. This new capability is the result of our on-going research with the Monterey Bay Aquarium Research Institute (MBARI)
Keywords
CCD image sensors; computer vision; mobile robots; navigation; path planning; remotely operated vehicles; underwater vehicles; user interfaces; video signal processing; autonomous underwater vehicles; hover-capable underwater vehicles; image-based user interface; mosaicing; navigation; ocean floor; underwater station-keeping; vision-based dead-reckoning system; Automotive engineering; Computer interfaces; Computer vision; Error correction; Navigation; Oceans; Remotely operated vehicles; Testing; Underwater vehicles; User interfaces;
fLanguage
English
Publisher
ieee
Conference_Titel
OCEANS '98 Conference Proceedings
Conference_Location
Nice
Print_ISBN
0-7803-5045-6
Type
conf
DOI
10.1109/OCEANS.1998.725761
Filename
725761
Link To Document