Ah, discussion is starting...
Concerning the computation of visible objects on a map, Tomoichi wrote:
> I am against it at present.
> Becasue, it can be calculated but it takes time to calucate 3D models.
> I am for it in future. When introducing 3D models, I would like
> to simulate evacuations and rescues in skyscrapers or underground
> malls.
For the time being we could have a simplified model that is calculated
offline, i.e. just once before the actual start of the simulation.
For example, the calculation would look like this:
for all pairs o1,o2 of static RCR objects (buildings, nodes, roads) that
are in less than 100m euclidean distance, compute whether there is an
obstruction (i.e. a building) between them. If not, add o1 to
Visible(o2) and vice-versa. Now, during simulation an agent standing on
location l would perceive all static objects in Visible(l) plus all
objects standing on locations in Visible(l). If we have a visibility
restriction like smoke it is easy to verify that online if the sets
Visibility(l) are precomputed.
The drawback of this approach is that we may allow too much to be seen,
e.g. several cars moving in line could all be seen. But I believe even
this could easily be checked online since the set of possible candidates
for occlusion is small.
Of course, real 3D visibility , esp. inside buildings, is another
matter. Our representation of the world is still too crude to reasonably
describe this kind of perception. That's why we propose to model it by a
function describing how long it takes to search buildings depending on
some variables known in the simulation.
michael
-- Michael Brenner Phone: (+49) (761) 203-8226 Institute of Computer Science Fax: (+49) (761) 203-8222 Albert-Ludwigs-University mailto:brenner@informatik.uni-freiburg.de Am Flughafen 17, Geb. 52 http://www.informatik.uni-freiburg.de/~brenner D-79110 Freiburg, Germany
This archive was generated by hypermail 2.1.3 : Mon 15 Nov 2004 - 14:55:37 GMT