Combined presentation of the following two papers: 
AUTHOR: 
TITLE: 
AUTHOR: 
TITLE: 
DISCUSSION: 
Elassal: 
Curry (USA): 
Sarjakoski: 
Curry: 
Sarjakoski: 
Curry: 
T. Sarjakoski (Finland) 
POTENTIAL OF EXPERT SYSTEM TECHNOLOGY FOR AERIAL 
TRIANGULATION 
T. Sarjakoski (Finland) 
SOFTWARE ENGINEERING IN PHOTOGRAMMETRY 
Thank you. This was very interesting. I'm sure 
more of you are in one way or another interested 
in the subject of artificial intelligence (AI). 
My organisation is starting a pilot program in AI. 
Our people are looking into the cartographic process 
trying to find areas which might be adoptable to 
AI solutions. I open the discussion to the floor. 
I wonder if you could give us a few more details on 
the actual system that you implemented. Did you use 
an object-oriented language for that implementation? 
No. The system in its current state is based only 
on Fortran. I should not say it, but that is the 
way it is. Because I didn't have the access to the 
right tools, as I mentioned in the presentation. 
I only used some AI algorithms and the expert system 
way of thinking. 
Then how far do you see the expert system descending 
into this eventual system? Do you imagine an actual 
bundle block adjustment program using expert system 
techniques to decide what algorithms to apply and 
use for data snooping? Or do you see an expert 
system shell simply calling existing high level 
language programs? 
It is fuzzy. I have one viewgraph here. Actually 
the situation is such that we have this kind of 
front/end solution with an analysis system. If we 
make this analysis system more intelligent to be a 
kind of expert system already then obviously the 
front/end expert system has to be less intelligent. 
From a user's point of view, it is critical how the 
system functions as a whole. 
My final question is, some people claim that large 
computational systems cannot be written in languages 
such as LISP with present LISP machines and so on. 
The stories about inefficiency are not any longer 
true. Do you believe that it is possible for a 
bundle block adjustment to be written in LISP? 
59