Distributed Collision Detection

Braddock Gaskill ([email protected])
Wed, 4 Oct 1995 23:15:54 -0400 (EDT)


Coming from a simulation background, collision detection is a very
computationally expensive process. Explicit FEA simulation codes often
spend the majority of their CPU time in the penetration detection
algorithm.

In a very large-scale multi-user virtual environment, it would
seem to me that it may be very worth-while to consider the distribution of
the collision detection task. Notification of a collision detection is,
after all, a relatively low-bandwidth message.

Perhaps it will help later if we presently consider collision detection
more as a "high-level" behavior or state of the VE that can be passed
down theough event notifications much the same way that an avatar
movement would be, then as a "low-level" mechanism which should
intrinsically be the task of the individual browser to impliment and handle.

-Braddock

----- http://pcil.ece.jhu.edu/~braddock ----- [email protected] -----


  • Next message: J Gwinner/VisNet, I: "testing VRML browsers"
  • Previous message: In Training: "Re: Object transfer"
  • Next in thesad: Mitra: "Re: Distributed Collision Detection"