Newsgroups: comp.robotics
Path: brunix!news.Brown.EDU!qt.cs.utexas.edu!cs.utexas.edu!zaphod.mps.ohio-state.edu!howland.reston.ans.net!gatech!concert!rock!taco!djbarnes
From: djbarnes@eos.ncsu.edu (DONALD JAMES BARNES)
Subject: Re: Run Rabbit...Run. Cat detectors.
Message-ID: <1993Feb23.163418.23891@ncsu.edu>
Originator: djbarnes@c00477-224wi.eos.ncsu.edu
Sender: news@ncsu.edu (USENET News System)
Reply-To: djbarnes@eos.ncsu.edu (DONALD JAMES BARNES)
Organization: North Carolina State University, Project Eos
References: <1mbht6INN2s9@clover.csv.warwick.ac.uk> <1mdeloINN7e3@roundup.crhc.uiuc.edu>
Date: Tue, 23 Feb 1993 16:34:18 GMT
Lines: 51


In article <1mdeloINN7e3@roundup.crhc.uiuc.edu>, cthang@stereo.csl.uiuc.edu (Thang Cao Nguyen) writes:
> esrbu@csv.warwick.ac.uk (Mr I M Hunter) writes:
> 
> >Dear all,
> >        I wish to build a small (tennis ball sized) mobile robot that has
> >a self-preservation instinct when faced with my cat. That is, it runs away
> >from it. The thing is how does it detect where the cat is and whether it
> >has pounced upon my poor, doomed robot. I was thinking of attatching an

> Hello,
> 
> Just off-hand, I got an idea: sense the electrostatic potential around
> the robot, and choose a direction along the gradient to run a way from:-)
> I think this will eliminate the need for your cute cat's carrying IR emitter.
> Furthermore, sensing electrostatic potential is probably "right": cats have
> their great fur coats with potentially lots of charge.  Also, voltage sensing
> is really straightforward.
> 
> To elaborate, just for fun, let's suppose the robot has 4 antenna: (a,b,c,d)
> 	(or 3, whatever)
> 
> 			    b |
> 			      I    c
> 			-----( )-----
> 			a     I
> 			      |d
> 
> one can built simple logic circuits (threshold, comparators and other
> analog functions) to process the measured potentials at a, b, c, and d, and 
> drive the 2 independent motors according to some sorts of strategy for 
> "running away".  For example: the robot can back away for a moment, then 
> turn and run in another direction, etc.  I'd suggest having some 
> time-constant RC circuits to add more flexibility to addjust the behaviors.
> 
> I expect that the robot could behave funny all by itself due to the 
> self-generated electrostatics, for example, while it's running on a carpet
> and the thresholds are low enough.  It may be fun to let it that way, in fact,
> or you can set a good threshold and make sure the robot statics are reall low.
> How to detect if the cat pounced upon the robot?  I don't know how.
> Any good way to "sense" if the antenna have been bent out of shape? :-)
> Probably too complex for a tiny robot to decide for itself.
> 
> 

You could even incorperate some microswitches on the antennas themselves to
detect getting hit or running into something.....


--Donnie

