Chuck
> 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 dsfine 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 write each of these behaviors in your system; including
> mapping the degrees of freedom 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 degrees of freedom in a model over time. I'm interested in how
> these systems deal with complementary control (i.e. the color and scale are
> independent degrees of freedom; 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)
>