Dwight Meglan (dwight++at++ht.com)
Fri, 3 May 1996 09:12:04 -0400
The application is a endoscopic simulator-- this is a small flexible shaft
camera that is used to explore and operate (through a working channel that
allows tools) in the body. We now have operating a simulator that allows us
to create models and move about inside of them using multibody dynamics for
the camera/shaft behavior and isects and collision reaction for constraint
inside the human anatomy/geometry.
We hand built a nested hierachy of the geometry using a combination of
Alias, wavefront, Designer's Workbench and some little custom inventor
utility programs. We found that the traditional modeling tools failed
miserably at times at preserving the hierarchy and definitely could not
create smoothed, textured models with the hierarchy. Here we built a flat
model, smoothed and textured it and then merged the normal and texture
coordinates into the hierarchical data files with the custom programs.
For isects, the very first time we do a first pass to find the geoset that
the camera segments are in -- note that the entire length of the camera
shaft must be checked each pass since is snakes through the anatomy and is
in contact in multiple places. Following geoset identification we do
primitive isects and compute reaction forces which are applied to the
multibody dynamics code (our own BTW). We record the geoset that the isect
occured in and next time start checking with it and then go to a full
hierarchy if this fails. In addition to isect checks, we built a void
geometry hierarchy which we check before doing isects. If the contact
segment is completely contained in one of the geosets of this hierarchy--
we're done, i.e. no isect, and we record this geoset for first check next
frame. Lastly, on an Onyx we run the isects on a separate processor. The
system works fine on an Impact also, just slow as you would expect.
I know that there are elaborate systems to do fast, large isect checks
(I-COLLIDE comes to mind) but I think this approach is a pretty reasonable
use of performer's built capabilities with a minimum of extra development.
The resulting simulation is quite nice-- the scope behavior is eerily
similar to the real situation.
Once again thanks for the input -- this mailing list is a tremendous
resource. I'm looking forward to meeting some of you at the developer's
forum.
--dwight
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dwight Meglan, PhD | Developers of complete surgery
simulation
Engineering Director | training systems and surgery simulation
High Techsplanations, Inc. | creation software tools
6001 Montrose Rd., Suite 902 |
Rockville, MD 20852-4874 | 301 984 3706 x238 301 984 2104 : FAX
dwight++at++ht.com | http://www.ht.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:52 PDT