Newsgroups: comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: TWO LEGGED ROBOT : PROJECT 'DEINONYCHUS'
Message-ID: <nagleCzMuLu.H2G@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <3aqd5b$bga@seralph9.essex.ac.uk>
Date: Mon, 21 Nov 1994 19:39:29 GMT
Lines: 24

kjwith@essex.ac.uk (K J Withers) writes:
>  This on its own is not that interesting ,BUT when the routine starts to 
>correct 
>the model the points A and C  rotate around the point B in such a manner that
>LIFT is created and the point B returns to the upright (although for a breif
>period as the routine over corrects). 
>So what is this LIFT, well in actual fact it is ANTI-GRAVITY.
>You don't beleive me?

     No, we don't.

     Writing a simulator for realistic physics of connected rigid bodies
is very difficult if you implement collision forces.  If you really
simulate rigid bodies, collisions involve infinite accelerations
applied over zero time with a finite energy product.  This leads to
unstable numeric computations.  There are ways around this, but none
of them are perfect.  Some implementatio approaches can introduce
huge energies and momentums during collisions for this reason, violating
conservation of energy and momemtum.  I've had considerable trouble with
this using good commercial simulators, and it's hard to get it right.

      I'd look for trouble in your simulation math.

					John Nagle
