92.03.08
========


From: tolman%asylum@hellgate.utah.edu (Kenneth Tolman)
Subject: SOC: Multitasked brains and robot bodies with VR %-)
Date: Sun, 8 Mar 92 18:44:10 -0700
Organization: University of Utah CS Dept



   One entirely in a virtual world would have ones body lying about while
the brain (hooked in through neurons) gamboled through cyberspace.  It
is possible that one could create the virtual world arena to appear and
interact exactly like the outer world.  However, this is doubtful, for it
seems to be a slow and unwieldly interaction we must deal with here.  For
instance, we must wait in lines to get money out of an ATM, we must
wait for people to pick up phones to talk to them, and diddle around waiting
for commercials to end on TV.

   A solution is to create autobots- programs that function for you that
do not require full attendance.  For example, an autobot could go and
perform a transaction at the bank for you.  A properly created autobot would
have an actualy MODEL of its owner, so it could interact on YOUR behalf.
For instance, in the finalization of a business deal, an autobot could know
its owners wishes and desires and negotiate within those bounds.  It could
know what you said typically to your parents, and answer the phone for you
and hold conversations!  You would tell it in advance perhaps "tell my parents
I will fly out on the 17th, or on the 21st according to my schedule."  Then
when your parents phoned, it would negotiate the day by discussing with them
your calender.  With a myriad of autobots running errands, one could 
accomplish many tasks at once.  When the autobot hit a snag (an unplanned
contingency) it would return control to you, so you would have the ultimate
say in all decisions.  This is effectively a distributed person- like a 
distributed multiprocessor doing many things at once- you can accomplish
many things all at the same time.  This is similar for a multiprocessor
switches processes during IO wait lag, which is the same thing that the
person does here.  They have the autobot doing all the IO and the critical
decisions fall back into the hands of the person (except the preprogrammed
decisions).  A person could accomplish many things at once, and with everyone
doing this it could possibly raise the overall productivity.

    However, why have the body lying around doing nothing?  Couldn't one
hook the body up to a computer, and have the body perform useful tasks?
Your body could clean your house and cook dinner while you were away in
virtual land, when you returned you could reenter and eat the food.  Also
one could have one's body exercise and weightlift!  So you would return
to a healthy, strong body faithfully lifting weights while your brain 
concentrated on business during the day.

   With a multitasked brain doing many things at once, you could end up having
to deal with yourself!  For instance, you may work at resolving problems
at the bank.  You send one of your autobots out to do some transaction with
the bank, and it eventually gets deferred to yourself.  Now the autobot
gets stuck and sends control back to owner, and you find yourself face to face
with yourself!  This is a simplistic scenerio, but it will happen.

   One alternative to having your body lift weights while you are away is
allow other people to use it as an end effector (or as their own body).  This
could run into some sticky situations fast, for the other person could kill
your body (and your brain residing therein) while you were away.  Having
robot (metal=unowned) bodies to use as end effectors is better.  These robot
end effectors would probably cost differently depending on demand in
certain regions.  For instance, moving into a body in France may be quite
expensive (for it is very popular) and operating a robot body in New York
quite cheap.  Supply and demand would end up pushing for new manufacture
distribution of robot bodies to popular areas.  Some regions may have caps,
such as wilderness areas only allowing 10 bodies within.  Here one may have
to share the experience with someone else, or have a lot of money.




From: thinman@netcom.com (Lance Norskog)
Subject: Re: TECH: Collision Prediction
Date: Sun, 8 Mar 92 12:29:25 PST



More from comp.graphics:

} From: eyal@echo.canberra.edu.au (Eyal Lebedinsky)
} Subject: Solution: shortest distance between two segment-paths.
} 
} [...]
} 
} The problem:
} 
} Given a line segment and a point(p) find the point(n) on the segment
} nearest the point(p). [ ... ]
} 
} By now I realized that my REAL need is to find the shortest distance
} between two line segments, not a line and a point. And moreover, as the
} lines represent the path of two points moving in a fixed rate in a
} straight line then I am only interested in the distance between two
} RELATED points on these paths. Strangely this problem was easier to
} see. Here is how I do it now:
} 
} Given a point a(x,y,z) and a segment from it va(x,y,z), a point b(x,y,z)
} and a segment from it vb(x,y,z) we observe that the parametric line
} segments are
}     (ax+d*vax,ay+d*vay,az+d*vaz) and (bx+d*vbx,by+d*vby,bz+d*vbz)
} Where d is between 0 and 1, and it is the SAME d for related points on
} the two line segments.
} 
} The distance (squared) is: 
}     (ax+d*vax-bx+d*vbx)^2 + (ay+d*vay-by+d*vby)^2 + (az+d*vaz-bz+d*vbz)^2
} 
} which expands to:
} 
}     d^2*((vax-vbx)^2 + (vay-vby)^2 + (vaz-vbz)^2) +
}     2*d*((ax-bx)*(vax-vbx) + (ay-by)*(vay-vby) + (az-bz)*(vaz-vbz)) +
}     (ax-bx)^2 + (ay-by)^2 + (az-bz)^2
} 
} The minimum is when the derivative is zero:
} 
}     2*d*((vax-vbx)^2 + (vay-vby)^2 + (vaz-vbz)^2) +
}     2*((ax-bx)*(vax-vbx) + (ay-by)*(vay-vby) + (az-bz)*(vaz-vbz)) == 0
} 
} Which is a straight forward calculation.
} 
} d = ((ax-bx)*(vax-vbx) + (ay-by)*(vay-vby) + (az-bz)*(vaz-vbz)) /
}     ((vax-vbx)^2 + (vay-vby)^2 + (vaz-vbz)^2)
} 
} However, this need not be calculated always. if the first term is
} negative then the nearest distance is between the starting points. If
} the first term is larger than the second then the end points are
} closest. Otherwise we need to calculate 'd' and subtitute into the
} parametric line squations.
} 
} In practice there are many common terms, so the math goes like this:
} 
} x  = bx-ax
} y  = by-ay
} z  = bz-az
} dx = vbx-vax
} dy = vby-vay
} dz = vbz-vaz
} 
} s = x*dx+y*dy+z*dz;
} if (s > 0) {
}       t = dx*dx+dy*dy+dz*dz;
}       if (s >= t) {
}               x += dx;
}               y += dy;
}               z += dz;
}       } else {
}               x += dx*s/t;
}               y += dy*s/t;
}               z += dz*s/t;
}       }
} }
} 
} neartest distance = x*x+y*y+z*z;
} 
} If all I want is to know when the two paths are closest then I can
} ignore the case (s >= t) because the end points will be the start
} points of the next time I do the test (next time interval). I may
} choose to ignore the case when (s < 0) as the objects are getting away
} from each other but for (s == 0) I have a real minimum.
} 

He didn't mention it before, but he appears to be working on 
collision detection also.

Lance Norskog
thinman@netcom.com




From: exv2447@ultb.rit.edu (E.X. Vanhensbergen )
Subject: TECH: Mailing List
Date: Sun, 8 Mar 1992 18:03:50 GMT
Organization: Rochester Institute of Technology



Does there exist some comprehensive mailing list
of all companies and research institutes working on
Research & Development of Virtual Reality type
systems?  If so, where could one obtain such a list?
Thanks,

-------------------------
Eric Van Hensbergen
Virtual Reality Research Group
Computer Science House
Rochester Institute of Technology
-------------------------

[MODERATOR'S NOTE:  The current AI EXPERT special has a list that was fairly
inclusive as of the end of 1991, when the booklet was put to bed.  But the
field is changing so quickly, any printed listing is bound to be out of date
as soon as it appears.  You might check the archives here for a relatively
good listing, but that is only so long as people continue to contribute
information for the archives.  Other leads would be appreciated! -- Bob
Jacobson]




From: swisher@cs.utexas.edu (Janet Marie Swisher)
Subject: CULTURE: After Battletech
Date: 8 Mar 1992 01:15:56 -0600
Organization: CS Dept, University of Texas at Austin



In article <1992Feb29.222455.12360@u.washington.edu> Peter Fraterdeus
<pfraterdeus@igc.org> writes:

>Regarding the Battletech Center debate...
>
>The most likely counter for "killing games" will be learning games.
>
>Education that challenges like a good old game of Pong (remember?), but
>enriches with rich references to vast encylopediae of accessable
>information....

Absolutely.  "Where in <whatever> is Carmen Sandiego?" is such a huge
hit, even spawning a TV show, that a VR version seems a logical next
step.  That would be right up the alley Peter describes.



