2.3 Available material
Before the room was closed, a small set of photos (less than 10)
was taken. The subject of these photos is restricted to a small set
of elements. They do not cover all the room and cannot be used
to guess an accurate geometry of the scene (see Figure 1).
Figure 1. Image of the room before closing.
A video of the room was also realised. It provides more
information than the photos, but some parts of the room were
not covered. Moreover, the image quality is quite poor and no
information is available on the camera.
Existing plans provide an estimate of the room geometry but are
doubtful on several points. They do not represent the as-built
state of the room. From the video, we know that many elements
are missing: pipes, filters, some tanks, etc. Drawings of the
electro-valves are also available but with no scale factor.
All this material (photo, video and plans) was combined a few
years ago, to build a 3D model of the room with an on-the-shelf
CAD modelling software (3Dstudio). According to the elements
used, this model gathers the collected information about most
parts and room topology but it is not geometrically reliable, and
cannot be used straight for an intervention in the room.
2.4 Special requirements on the model
The expected model should fulfil mission specific requirements.
For instance, one of the first tasks of the decommissioning work
is to drill a hole in the ceiling above one of the tanks to latter
check the content and empty this tank using robotics means. An
obstacle-free space to reach the tank and the corresponding
position where to drill the ceiling should be determined from the
acquired model.
3. EQUIPMENT SELECTED
3.1 Camera
In order to build a 3D model, we decided to drive a sensor in the
room. We chose video as a good compromise between the
amount of information provided and the limited cost of the
solution. CCD sensor was acceptable since we had to work for a
limited amount of time. Indeed the chosen sensor worked for
several months with no major problem.
We chose a small size 17mm diameter endoscopic camera. In
order to reduce exposure to radiations and to limit heat brought
into the room, the electronics of the camera was deported off the
room. A 768 x 576 pixels resolution was retained for cost and
size considerations.
3.2 Lighting
As the room was completely dark, we needed to bring some
light. We did not want to use low light level camera for
problems of cost and because of the bigger size of these kinds of
sensors. For reasons of safety, we had to avoid bring heat inside
the room. We chose fibre optic light instead. Several optic fibres
were set around the camera lens to provide an even lighting.
The diameter of the set camera and optic fibre ring is less than
25mm (Figure 2).
Figure 2. Camera and optic fibre lighting ring
We made several tries in a dark room with a high reflectivity
stainless steel parts to make sure of this solution.
3.3 Lens
We paid a special care on lens choice. We had these constraints:
e We wanted to have a very good knowledge of focal
lens and chose a fix focal length
e We wanted images with minimum distortions and
therefore had rather use long focal length.
e We needed to see 1-meter wide objects 1.5 meters
away from the camera.
As a compromise between these constraints, we chose a 7.5 mm
lens for our % inch CCD camera.
3.4 Carrier
To carry the camera, a 3 degrees-of-freedom articulated mast
was built. The joints (Z translation, pitch and yaw) were
equipped with sensors to ensure a 1 mm measurement accuracy
of the position of the camera (see Figure 3).
The range of the joints were the following:
e Zaxis: 1500 mm
e Pitch: 360 degrees
e Yaw: 150 degrees
Figure 3. Carrier with camera and ring lighting
—176—
The
unkı
To €
pho!
sele:
indi:
cam
com
viev
hole
Inte
viev
199
Gra
relic
reli:
recc
Hra
CA
deri
The
kno
and
The
198
des
he :
soli
theı
and
the
pro
20€
(Ev
ori
Ear
sof
frai
tas]
eitl
pro
wit
im]
res
its
uni
In
cot
for
all
car
pat
(fx
pai
tra