was free from obstacles did not give much information on the 
objects in the environment. 
6.3 Remarks on lighting 
Ring light was satisfying in the area of interest and did not make 
any problem. The lighting was though very weak 3 meters away 
from the camera. 
7. ROOM MODELLING 
All the preceding remarks made important the use of hypothesis 
on the room structure, while modelling. 
7.1 Heat exchanger 
The heat exchanger was the first object modelled in order to put 
in evidence the contact of the arm with this obstacle. Once 
obtained, this model could be used on-line to control the camera 
motions and avoid collisions against the flanges lying around 
the heat exchanger. 
      
Figure 4. Carrier arm in the hole and heat exchanger 
7.2 Stainless steel tanks 
Three of the tanks were supposed to have identical dimensions. 
Pyramide confirmed this, even if one of the tanks could not be 
seen entirely from a single viewpoint. Pyramide led to re- 
estimate the diameter of tank cover to respect the proportions of 
the model in the image. It is quite admissible that these parts 
realisation has not exactly respected the initial plans for 
convenience reason. Pipes coming out of the tank were 
supposed to be orthogonal to the cover. This hypothesis allowed 
compensating for the wrong position measurement of the arm. 
     
    
Figure 5. Image of tanks B and A and image of tank C 
Because of the limits on the arm translation, we could not get 
images of the bottom and of the feet of the tanks. The 
dimensions of the feet were guessed from floor modelling. 
  
Figure 6. Model of tank B 
7.3 Electro-valves 
  
Figure 7. Electro-valves and pipes 
Superimposed on video images, the initial model of the electro- 
valves appeared incorrect. The plans used to build this model 
had no scale factor and they appeared to be incomplete, leading 
to some ambiguous 3D interpretation. The positions guessed in 
the initial model were also wrong. Furthermore, electro-valves 
include a mobile part that did not appear in the plans. A model 
of the electro-valves was completely redesigned with Pyramide. 
Some little pipes linked to the electro-valves not present in the 
initial model were modelled with Pyramide. They needed to 
figure in the model since they can represent some obstacles for 
the future task in this area. 
   
Figure 8. Model of the electro-valves from top and front. 
7.4 Filters, waste tanks, retention tank 
For distant objects, we used the same assumptions for 
modelling than the ones used to build the initial model because 
of the inaccuracy of arm position measurement. This was 
required for the two waste tanks, a retention cylinder and filters 
on the wall. 
—178— 
   
o 4 = 
gG'OG. e 
ON pT a mi