[comp.graphics] Luxo lamp jumps hurdles

v134kkut@ubvmsb.cc.buffalo.edu (David W Tinklepaugh) (10/21/90)

first, let me say this article is directed more towards computer visual
3D simulation than computergraphics in general...

I read this fascinating article once but can't remember where.  I was
wondering if someone could give me a reference for it or information on 
something similar:

ok, these guys were animating a luxo lamp on a computer.  they gave each joint
a certain amount of maximum strength.  told it it had to jump over an invisible 
hurdle placed a certain height a certain distance in front of it.  Assuming 
natural laws of physics (gravity, etc.).   The lamp would crouch in a certain
way and, jump and maneuver itself as it flew through the air to go over the
hurdle.  the more odd the placement of the hurdle, the more oddly the lamp
jumped.  until the guys told  the computer 'jump', they had no idea what it would
do.  

also, what is this called?  parameterization?  goal orientated programing? 
object/environment interaction?  

something similar to this is simulating a chain hanging vertically, swaying in
the 'wind' or two tops spinning, hitting each other.   The person specifies 
starting conditions and the rest of the action is calculated from a set of
physical laws of their interaction.   


       ______________________________________________________________________
      /                               /                              \      /
     /    -Dave-   U. of Buffalo     /    Wrap it up, I'll take it!   \    /
    /   v134kkut@ubvmsa (bitnet)    /      -Fabulous Thunderbirds      \  /
   /_______________________________/____________________________________\/

foo@titan.rice.edu (Mark Hall) (10/22/90)

In article <41853@eerie.acsu.Buffalo.EDU> v134kkut@ubvmsb.cc.buffalo.edu writes:
)ok, these guys were animating a luxo lamp on a computer.  

   See "Spacetime Constraints", Andrew Witkin and Michael Kass, 
 in Computer Graphics, Volume 22, number 4, PP. 159-168. 

)something similar to this is simulating a chain hanging vertically, swaying in
)the 'wind' or two tops spinning, hitting each other.   

   See "A Model based on Dynamic Constraints", Ronen Barzel and Alan Barr,
 in  Computer Graphics, Volume 22, number 4, PP. 179-188.

   See also 
	``Rigid Bodies Dynamics Simulation'' [videotape]
	SIGGRAPH Video Review  Number 28 (1987).
	ACM SIGGRAPH, New York, segment 11.

   and
	``Caltech Studies in Modeling and Motion'' [videotape]
	SIGGRAPH Video Review Number 28 (1987).
	ACM SIGGRAPH, New York, segment 3.



  - mark

nagle@well.sf.ca.us (John Nagle) (10/23/90)

v134kkut@ubvmsb.cc.buffalo.edu (David W Tinklepaugh) writes:

>I read this fascinating article once, but can't remember where.  
>These guys were animating a Luxo lamp on a computer.  They gave each joint
>a maximum strength; told it it had to jump over an invisible 
>hurdle placed a certain height a certain distance in front of it.  ...

     The article is "Spacetime Constraints", by Andrew Witkin and Michael
Kass, and it appeared in ACM Trans. on Computer Graphics, Vol 22, #4, 
August 1988.  

>What is this called?  parameterization?  goal orientated programing? 
>object/environment interaction?  

     It's called nonlinear optimization.  This is a very powerful
technique, but it requires extensive CPU resources.  Kass's first
implementation took 20 minutes on a Symbolics 3600 to compute a simple
jump.  A later implementation, which used Mathematica to differentiate
and simplify the functions involved so that C code could be generated and
linked into the optimizer, required only 20 seconds on a 25 MIPS machine.

     I encourage people to read up on this technique and pick up from
where these people left off.  The work was going well, but Schlumberger
decided to get out of the AI business and closed its Palo Alto lab,
where this work was done.  Witkin is now at CMU, and Kass is at Apple.

     If someone plans serious work in this area, please let me know.

					John Nagle

pjyamamoto@watdragon.waterloo.edu (Peter Yamamoto) (11/03/90)

In article <41853@eerie.acsu.Buffalo.EDU> v134kkut@ubvmsb.cc.buffalo.edu writes:
>first, let me say this article is directed more towards computer visual
>3D simulation than computergraphics in general...
>
>I read this fascinating article once but can't remember where.  I was
>wondering if someone could give me a reference for it or information on 
>something similar:
>
>ok, these guys were animating a luxo lamp on a computer.  they gave each joint
>a certain amount of maximum strength.  told it it had to jump over an invisible
>hurdle placed a certain height a certain distance in front of it.  Assuming 
>natural laws of physics (gravity, etc.).   The lamp would crouch in a certain
>way and, jump and maneuver itself as it flew through the air to go over the
>hurdle.  the more odd the placement of the hurdle, the more oddly the lamp
>jumped.  until the guys told  the computer 'jump', they had no idea what it would
>do.  

Several people have given references to the original literature but you may
also be interested in:

Michiel van de Panne, Eugene Fiume, and Zvonko Vranesic,
Reusable motion synthesis using state-space controllers,
Computer Graphics, Voulme 24, Number 4, August, 1990

Ie from the SIGGRAPH '90 conference proceedings.

This is work done at the University of Toronto, Cananada.

Peter
-- 
:-><><><><><><><><><><><><><><><><  YAMO!-)  ><><><><><><><><><><><><><><><><-:
pjyamamoto@watsol.waterloo.edu      /\     University of Waterloo
          @watsol.waterloo.cdn     /\/\    Department of Computer Science
          @watsol.uwaterloo.ca    /\/\/\   Waterloo, Ontario N2L 3G1 CANADA