Newsgroups: comp.robotics,comp.ai.genetics
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!howland.reston.ans.net!ix.netcom.com!netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: How does a six-legged animat turn/rotate?
Message-ID: <nagleD4xIsv.K61@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <Pine.ULT.3.91.950303014905.25482B-100000@rac5.wam.umd.edu>
Date: Sat, 4 Mar 1995 18:46:07 GMT
Lines: 26
Sender: nagle@netcom14.netcom.com

Keith Wiley <keithw@wam.umd.edu> writes:
>This seems like a fairly simple question but actually, I can't come to a 
>definite conclusion.  I have finished the preliminary stages of my 
>walking robot program.  I exists in three dimensional space using X,Y,Z 
>coordinates to define the endpoints of leg and body segments.  At the 
>moment I have a six legged creature with double-segmented legs that uses 
>bottom-up processing to walk.  This walking algorithm needs a lot of work 
>(it's fairly program defined at this point as opposed to truely emergent) 
>but I don't know for sure how a six legged animat rotates.  Does it stick 
>various legs under it and opposite legs further away from it and then 
>change the angles so the legs are equally distributed again or do 
>segmented insects pick up their legs and rotate like a hinge between 
>segments and put their legs back down.  There are a couple of ways I 
>could do this I suppose but mimicking nature always seems like the best 
>bet and I don't know what it is that nature does for sure.
>Thanks for any help.

      Use force control.  Put the horizontal motion of all legs in force
control mode, then push forward with the legs on one side and backward
on the other.  If a leg breaks free of the ground, cut the force in half
for all legs.  If a leg reaches its limit, pick it up and re-place it.

      Take a look at Randall Beer's book for a neural architecture that
does something like this.

						John Nagle
