Navigator
Navigation
|, pp. 722-
[IT Press,
of Kalman
tering. In:
n filtering,
Aerospace
on by the
-486.
1 filtering,
Heidelberg
GPS/INS
+ IEEE
6.
€ Kalman
3.203.
| GPS/INS
ience and
echnology
339.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
DIRECT GEOREFERENCING WITH ON BOARD NAVIGATION COMPONENTS OF
LIGHT WEIGHT UAV PLATFORMS
Norbert Pfeifer*, Philipp Glira*, Christian Briese^^
* Institute of Photogrammetry and Remote Sensing of the Vienna University of Technology, Austria - (np, pg, cb) @ipf.tuwien.ac.at
® LBI for Archaeological Prospection and Virtual Archeology, Vienna, Austria
KEY WORDS: direct georeferencing, UAV, orientation, GPS/INS, IMU, photogrammetry
ABSTRACT:
Unmanned aerial vehicles (UAV) are a promising platform for close range airborne photogrammetry. Next to the possibility of carrying
certain sensor equipment, different on board navigation components may be integrated. These devices are getting, due to recent
developments in the field of electronics, smaller and smaller and are easily affordable. Therefore, UAV platforms are nowadays often
equipped with several navigation devices in order to support the remote control of a UAV. Furthermore, these devices allow an automated
flight mode that allows to systematically sense a certain area or object of interest. However, next to their support for the UAV navigation
they allow the direct georeferencing of synchronised sensor data.
This paper introduces the direct georeferencing of airborne UAV images with a low cost solution based on a quadrocopter. The
system is equipped with a Global Navigation Satellite System (GNSS), an Inertial Measurement Unit (IMU), an air pressure sensor, a
magnetometer, and a small compact camera. A challenge using light weight consumer-grade sensors is the acquisition of high quality
images with respect to brightness and sharpness. It is demonstrated that an appropriate solution for data synchronisation and data
processing allows a direct georeferencing of the acquired images with a precision below 1 m in each coordinate. The precision for roll
and pitch is below 1 ° and for the yaw it is 2.5 °. The evaluation is based on image positions estimated based on the on board sensors
and compared to an independent bundle block adjustment of the images.
1 INTRODUCTION
Unmanned aerial vehicles (UAVs) are promising platforms for
the provision of geo-referenced Earth Observation(s) for a num-
ber of reasons: ease of deployment, costs, and close range ac-
quisition from an elevated position, more specifically ‘airborne
vertical close range photogrammetry’. However, also oblique
and horizontal viewing at large scale from elevated positions are
an option. The advantages materialize in comparison to manned
aerial vehicles, on the one hand, and to terrestrial elevated plat-
forms, e.g., ladders, on the other hand. These advantages, how-
ever, only apply to relative light-weight UAVs, which are there-
fore restricted with respect to payload and therefore sensor qual-
ity.
The restriction of payload advocates for a careful design of the
entire system, comprising, next to the platform itself and the imag-
ing sensor also a position and orientation (POS) component and
the power supply for the sensors and the aerial vehicle. For the
purpose of navigation, e.g., flying along pre-defined waypoints,
or taking user control input in the form of movement direction
and speed, UAVs are typically equipped with position and mo-
tion sensors. Medium weight platforms, e.g., with a payload of
15kg, can accommodate a high quality inertial navigation sys-
tem (INS) and on board storage, to compute the vehicle’s trajec-
tory by Kalman Filtering in post-processing. For small platforms
the accelerations and rotation rates are typically measured with
micro-electro-mechanical systems (MEMS).
In this contribution we demonstrate for a light weight UAV, a
quadrocopter with platform and payload together below 1 kg, that
the on board components for navigation can be used for direct
georeferencing of the acquired imagery.
This has the advantage of having only the cameras and its mount-
ing as additional payload, and only the existing GNSS, gyro-
scopes, accelerometers, air pressure sensor, and magnetometer
are used for direct georeferencing. Specifically, we provide
487
e information on the assembly for full exploitation of the qual-
ity, that can be delivered by the camera,
e the methods necessary for synchronizing and processing the
data streams of the POS and image sensors,
e the quality obtained by the above procedures, specified as
accuracy of all elements of the exterior orientation of the
acquired images.
Concentration is laid on direct geo-referencing, i.e. also no ex-
ploitation of tie points is performed (integrated geo-referencing),
which has the advantages and drawbacks as given in (Cramer et
al., 2000).
In the remainder of the introduction related work is reviewed. For
general information on UAVs the reader is refered to (EisenbeiD,
2009) and (Everaerts, 2008). In Sec. 2 the UAV and the sen-
sors will be described. Following, in Sec. 3, the methods for the
computation of the trajectory are given, including information on
the synchronization. The method of evaluation and the obtained
quality are given in Sec. 4.
1.4 Related Work
Direct georeferencing of UAVs, if not performed with high grade
INS and differential GPS, was investigated to some extent before.
Direct georeferencing can be performed with GPS and INS, but
alternatives are tracking by tacheometers.
(Eisenbeif et al., 2009) investigated the accuracy of the trajec-
tory determined with low cost GPS receivers onboard a Survey-
copter 1B. A 360^ prism was mounted on the UAV and its 3D
position was measured with a tracking total station. The differ-
ences between direct georeferencing, using a Kalman-Filter for
the integration of GPS and INS, has a std.dev. in (X, Y, Z) of
of 70 cm, 40 cm, and less than 20 cm, the offset reach up to 2 m.
Because of the superior accuracy of the tracking total station, this
