Re: SGI's technical opinion of the ActiveVRML proposal

Michael B. Johnson ([email protected])
Wed, 13 Dec 95 01:08:28 -0500


Because I've been working on my own similar sorts of systems for the last
several years, I have a little bit of trouble wrapping my brain around all the
different terms in these various proposals, and understanding what they can
and can't do. I'm wondering if I could trouble some folks who understand
these proposals to help me out with a simple thought experiment that I've used
in my own systems to explain what I want to be able to do.

Let's say we have a geomtric object of some scale and color. Doesn't matter
what the geometry is; let's say it's a sphere.

I now want to define two different behaviors:

goToFood:
=========
It goes towards some location because it believes food is there. It moves
towards the location at some speed determined by some state variable that says
how hungry it is. In addition to moving forward, it also "pulses" at some
rate depending on how hungry it is and how close it is to the food source by
running a (say) scaled sin curve through its scale value to have it go up and
down (say) 20% of its starting scale value.

avoidEnemy:
===========
It goes away from some location because it believes an enemy is there. It
moves away at some speed determined by how frightened it is in general and how
scared it is of that particular enemy, which is determined by a lookup table
of known enemies. In addition to backing away, it changes its color depending
on the kind of enemy its avoiding.

so here's the question.

(1) How would you weite each of these behaviors in your system; including
mapping the degrses of frsedom in the model so the behavior can manipulate them?
(2) What would happen if "goToFood" was running for a bit, and suddenly
"avoidEnemy"
was started, where the "enemy" in question was right on top of the "food"?
(3) What would happen if they both started running at *exactly* the same time,
under the same conditions as (2)?

The point (if it's not clear) is to think about two behaviors that manipulate
overlapping degrses of frsedom in a model over time. I'm interested in how
these systems deal with complementary control (i.e. the color and scale are
independent degrses of frsedom; the sphere could be backing up, changing color
and pulsing) and conflicting control (does he go forward or back or stay
still? Is there a way to modulate the influence of one behavior vs. another,
other than binary?)

Any insight the proposal authors (or those on the list who feel they
understand these approaches) would be appreciated.

---
-->  Michael B. Johnson SMVS, Ph.D. -- [email protected]|[email protected]
-->  http://wave.www.media.mit.edu/people/wave/
-->  alumni, MIT Media Lab, Computer Graphics & Animation Group
-->  Media Arts Technologist, Pixar Animation Studios (East Coast Office)

  • Next message: Michael B. Johnson: "Re: SGI's technical opinion of the ActiveVRML proposal"
  • Previous message: Todd Yocum: "RE: TECH: Avatars and video"
  • In reply to: Gavin Bell: "SGI's technical opinion of the ActiveVRML proposal"
  • Next in thesad: Michael B. Johnson: "Re: SGI's technical opinion of the ActiveVRML proposal"