PHOTOGRAMMETRY-DERIVED NAVIGATION PARAMETERS FOR INS
KALMAN FILTER UPDATES
F. A. Bayoud, J. Skaloud, B. Merminod
Geodetic Engineering Laboratory (TOPO), Swiss Federal Institute of Technology/Lausanne (EPFL)
Bât. GR, CH 1015, Switzerland — (fadiatef. bayoud, jan.skaloud, bertrand.merminod)@epfl.ch
Commission V, WG V/2
KEY WORDS: Mobile Mapping; Navigation; Robotics; Integration; IMU/INS; Kalman Filter.
ABSTRACT:
In this paper, a GPS-independent mobile navigation and mapping system is introduced. This system employs the
photogrammetric intersection to determine the coordinates of the surrounding objects and then it uses these newly
known objects to compute its own position, when it moves, by the photogrammetric resection. The photogrammetric
resection output — the exterior orientation parameters — is then used as external measurements in an INS Kalman Filer
to compute the filtered exterior orientation parameters to do the intersection. This methodology of navigation and
mapping is applied in the robotics community using different sensors and methods, where it is called Simultaneous
Localisation And Mapping (SLAM); SLAM is the problem of mapping an area and at the same time using the map to
locate the robot. In this paper, the solution for SLAM by integrating photogrammetry resection output and INS via a
Kalman Filer is described.
1. INTRODUCTION
It is fundamental that mobile mapping systems use
two independent components; one is for positioning,
and the other is for mapping. Looking closely at the
mapping component, one can find that it might also
be used for positioning. This idea is extensively used
in the robotics community.
Robotics Simultaneous Localisation And Mapping
(SLAM) is the problem of mapping the environment
surrounding the robot and at the same time using this
map to determine the location of the robot (Csorba.
1997; Newman, 1999). Currently, indoor robotics
SLAM is solved using LASER scanners to locate the
robot relative to the structured environment and at the
same time to map this environment. LASER scanners
showed to be a very good tool where the accuracy of
localisation is within the centimetre level for short
distances and favourable geometry. Outdoors robotics
SLAM is not feasible with LASER scanners due to
the absence of simple geometric environment.
Therefore, different tools have to be used.
Recently, the use of visual methods, integrated with
Inertial Measurement Units (IMU), has gained
interest. These visual methods rely on one or more
cameras (or video). Yet, no clear indication about the
mapping methods and integration algorithms is
clearly illustrated in the relevant literature (Journal of
Robotics Systems; V. 21; Issues 1 and 2).
In Geomatics Engineering, navigation and mapping
have been two central research topics for decades. A
classical tool for mapping is Photogrammetry, where
sequence of stereo-images are captured and analysed.
As for navigation systems, the coupling of inertial
sensors and Global Navigation Satellite Systems is
indispensable. In addition, photogrammetric resection
can also be used for positioning.
In this paper, a mobile mapping system is introduced
— independent of the GPS/INS integration - by
employing two CCD cameras and one IMU (Figure
1). The two CCDs will be used for mapping through
photogrammetric intersection, and for positioning
through photogrammetric resection. The presence of
an IMU is significant in cases where images cannot
be used for reasons of visibility and when certain
manoeuvres do not guarantee knowledge of the
environment; e.g. when the robot or vehicle captures
images of an unknown environment.
Thus, The relation. between photogrammetry and
SLAM is, analysed in this study. This relation has not
gained much attention due to the fact that SLAM
requires full automation, which until recently was far
from photogrammetry. Many attempts were directed
towards the full automation of photogrammetry. This
goal has been still falling short due to the need of
high level of artificial intelligence. Having in mind
that it is only a matter of time to reach full
automation, an investigation on SLAM from the
Geomatics point of view using navigation and
mapping modus operandi is essential.
cop «c Cup
EL
mL. E
e 2 o [5] I £ o9 e 2
Vehicle
Figure 1: The system
Interna