Hindawi Publishing Corporation EURASIP Journal on Advances in Signal Processing Volume 2009, Article ID 387308, 18 pages doi:10.1155/2009/387308 Research Article Vision-Based Unmanned Aerial Vehicle Navigation Using Geo-Referenced Information Gianpaolo Conte and Patrick Doherty Artificial Intelligence and Integrated Computer System Division, Department of Computer and Information Science, Link¨ oping University, 58183 Link¨ oping, Sweden Correspondence should be addressed to Gianpaolo Conte, giaco@ida.liu.se Received 31 July 2008; Revised 27 November 2008; Accepted 23 April 2009 Recommended by Matthijs Spaan This paper investigates the possibility of augmenting an Unmanned Aerial Vehicle (UAV) navigation system with a passive video camera in order to cope with long-term GPS outages. The paper proposes a vision-based navigation architecture which combines inertial sensors, visual odometry, and registration of the on-board video to a geo-referenced aerial image. The vision-aided navigation system developed is capable of providing high-rate and drift-free state estimation for UAV autonomous navigation without the GPS system. Due to the use of image-to-map registration for absolute position calculation, drift-free position performance depends on the structural characteristics of the terrain. Experimental evaluation of the approach based on offline flight data is provided. In addition the architecture proposed has been implemented on-board an experimental UAV helicopter platform and tested during vision-based autonomous flights. Copyright © 2009 G. Conte and P. Doherty. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1. Introduction One of the main concerns which prevents the use of UAV systems in populated areas is the safety issue. State-of-the- art UAVs are still not able to guarantee an acceptable level of safety to convince aviation authorities to authorize the use of such a system in populated areas (except in rare cases such as war zones). There are several problems which have to be solved before unmanned aircraft can be introduced into civilian airspace. One of them is GPS vulnerability [1]. Autonomous unmanned aerial vehicles usually rely on a GPS position signal which, combined with inertial mea- surement unit (IMU) data, provide high-rate and drift-free state estimation suitable for control purposes. Small UAVs are usually equipped with low-performance IMUs due to their limited payload capabilities. In such platforms, the loss of GPS signal even for a few seconds can be catastrophic due to the high drift rate of the IMU installed on-board. The GPS signal becomes unreliable when operating close to obstacles due to multipath reflections. In addition, jamming of GPS has arisen as a major concern for users due to the availability of GPS jamming technology on the market. Therefore UAVs which rely blindly on a GPS signal are quite vulnerable to malicious actions. For this reason, this paper proposes a navigation system which can cope with GPS outages. The approach presented fuses information from inertial sensors with information from a vision system based on a passive monocular video camera. The vision system replaces the GPS signal combining position information from visual odome- try and geo-referenced imagery. Geo-referenced satellite or aerial images must be available on-board UAV beforehand or downloaded in flight. The growing availability of high- resolution satellite images (e.g., provided by Google Earth) makes this topic very interesting and timely. The vision-based architecture developed is depicted in Figure 2 and is composed of an error dynamics Kalman filter (KF) that estimates the navigation errors of the INS and a separate Bayesian filter named point-mass filter (PMF) [2] which estimates the absolute position of the UAV on the horizontal plane fusing together visual odometry and image registration information. The 2D position estimated from the PMF, together with barometric altitude information obtained from an on-board barometer, is used as position measurement to update the KF.