Path: utzoo!utgpu!news-server.csri.toronto.edu!ephemeral.ai.toronto.edu!ai.toronto.edu!green
Newsgroups: comp.robotics
From: green@ai.toronto.edu (Anthony Thomas Green)
Subject: Re: HackerBot Information
Message-ID: <90Jul31.180211edt.8310@ephemeral.ai.toronto.edu>
Keywords: ubiquitous movit robot kits inexpensive but dumb
Organization: Department of Computer Science, University of Toronto
References: <1990Jul30.004821.27254@talos.pm.com> <5974@jhunix.HCF.JHU.EDU>
Distribution: na
Date: 31 Jul 90 22:02:26 GMT
Lines: 42

Ah yes, the MOVIT. Here's a fun project a friend and I did last year:

The MOVIT robot has a powerful (:-) 4 bit controller which we promptly
disconnected from the motors and removed. We then interfaced a 6809
to the motors and wrote a LOGO-like interpreter in Concurrent Euclid.
The robot could perform all of our LOGO commands, with one nifty twist.
We wrote an ON-KEY-DO command which would map key strokes on to 
LOGO procedures. A keyboard monitoring task would watch for these keys
and inform the interpreting task, which would push what it was doing
on to a stack and jump to the requested procedure. 

For instance,

ON 'T' DO TWIST_AND_MOVE

TO SQUARE (SIDE)
	REPEAT 4
		FORWARD SIDE
		RIGHT 90
	END REPEAT
END

TO TWIST_AND_MOVE
	RIGHT 20
	FORWARD 20
	REVERSE 20
	LEFT 20
END


Now if we ran SQUARE 100, and hit 'T' a bunch of times, the interpreter would
nest all of the TWIST_AND_MOVE procedures, causing the robot to do a little
jig. Eventually it would finish everything up and sit where it began.
The system was clever enough to know that if it was interrupted in the
middle of a FORWARD 100, by an ON-KEY-DO, it would finish it up when it
returned.

Anthony T. Green
green@ai.toronto.edu