Alt-BEAM Archive

Message #13598



To: beam@sgiblab.sgi.com
From: turtletek@aol.com
Date: Mon, 17 Apr 2000 20:07:21 EDT
Subject: Re: Photopopper tactile sensors...


In a message dated 4/17/00 6:34:15 PM Central Daylight Time, gabe@mediax.com
writes:

> I've finished a phtotopopper recently (ala chiu style). I'd like to put
> tactile sensors on it. I think I have a method that is
> easier to implement than any of the designs I've seen. Use standard
tactile
> sensors and have them simply short the photodiode on the
> appropriate side. I tested this on my current photopopper and it seemed to
> work fine. Any feedback?
>

Are you sure they work fine? Do remember that the standard configuration
allows the bot to totally ignore the tactile switches if both are triggered
at the same time. That way the bot at least doesn't freeze up. Does your
configuration provide this feature?

-Brien the TurtleTek
www.extremesc.com/turtletek



13599 Tue, 18 Apr 2000 00:29:22 GMT Re: Try building a robot that tries to avoid getting stepped on. beam@sgiblab.sgi.com "Justin Fisher"
>As a result, when the Deadly Darkness approached from above, the robot
>might zoom straight forward, spin left or right in place until it found a
> "good" place to run and then scoot or even simply zoom straight backwards.

I've built a robot with similar functionality, but using two SE's instead
of logic - a normal inverter-based SE overlayed with a D1. The D1 would
normally not affect the operation, but when it was triggered by a lower
light level, it would run to a lower voltage than the other SE before
switching off again, thus the bot always had a sizeable "panic" reserve in
its 2.5F cap that was triggered by the approach of the Deadly Darkness.
The problem was that setting the D1 to the sensitivity required to notice
the shadow (in an otherwise very bright environment) caused serious
charging losses. I might have tweaked or replaced it (I now have a far more
efficient design, using err.. logic :-), but as one of my earlier bots, it
also had some mechanical shortfalls so it seemed more exciting to just
start another bot :-)



13600 Mon, 17 Apr 2000 17:35:56 -0700 RE: Try building a robot that tries to avoid getting stepped on. "'beam@sgiblab.sgi.com'" Wilf Rigter Hi John,

Thanks for the Johuco URL.

I sure like the feel of this site, the products and what they have to say.

The GRAK concept is closest to my own idea of a CPUless robot controller
(soccer anyone?). It uses the same GRAK principle but I hardwire behaviours
in a 1M EPROM which are expressed by various combinations of sensors and
behaviour selectors (up to 14 function generator inputs) together with the
sequence count (3 bits) applied to the address lines. The 8 bit data bus
output can be used to drive up to 8 motors (with drivers) or combinations of
motors and other output devices. The logic would require as little as just 2
chips. Adding feedback (from latched outputs) can turn this into a more
complex state machine. And a CMOS EPROM requires only a few milliamps of
current when active so it's solar compatible.

regards

wilf

> -----Original Message-----
> From: John A. deVries II [SMTP:zozzles@lanl.gov]
> Sent: Monday, April 17, 2000 4:01 PM
> To: beam@sgiblab.sgi.com
> Subject: Try building a robot that tries to avoid getting stepped on.
>
> Was Mr. Robinson's challenge. Ben Hitchcock replied:
>
> >Female chickens will stoop down whenever a shadow falls over them,
> because
> >they think that a rooster is about to mate with them. A shadow cast by
> a
> >human works just as well as a rooster. I wonder just how 'sure' a robot
> >needs to be?
>
> His last sentence, I think, is a real clue to the situation. Wandering
> off
> thinking about various technologies is not necessarily the right place to
> start with this challenge.
>
> Firstly, there isn't a really good definition of "environment", "avoid",
> or, for that matter, "being stepped on". Granted, everyone might rush to
> say that at least the latter is obvious -- the robot should not be stepped
>
> on by a person. If you didn't want cockroaches stepping on your robot, or
>
> your family cat, or the elephant in your back yard you've got a huge range
>
> of scale. Be that as it may, let's say that the robot is supposed to
> behave in such a manner as to avoid being stepped on by a person.
>
> Well, what does that take? What behaviors are involved? The aggressive
> manner has been suggested but actually fails -- so what if someone gets
> lots of needles in their foot (silly person!)? They still have stepped on
>
> the robot.
>
> I think that the behavior that best fits the situation would be for the
> robot simply to Not Be There when a foot comes down. The truly bizarre
> imagination might think of "BEAMing" the robot (sorry, just couldn't avoid
>
> the pun) somewhere else, but we don't know how to do that yet. Thus we
> are
> left with physical motion, using legs or wheels or whatever. In any case,
>
> the robot is more-or-less forced to stay on the surface (of wherever it
> is)
> unless it has a decent early warning system and could fly away before the
> foot actually got close. However, this "closeness" would be hard to
> define
> -- just as Ben writes, "how 'sure' a robot needs to be"? Generally
> speaking, the early warning method would cause the robot to flee from ANY
> change in its environment presuming it could figure out the direction of
> the change and this is pretty wasteful of energy.
>
> Jonathan Connell, who was Rodney Brook's student something like twelve
> years ago, came up with something he called a "photovore" (a teensy bit
> earlier than the BEAM phenomenon, I think.) You can wander off to his
> site
> http://www.pcrealm.net/~johuco/photov.html and buy an entire kit; you
> could
> also read about Richard Hughes' modification of the device
> http://members.aol.com/rich924/html/vore.html. The cool thing about this
> photovore is that it had a light sensor pointing _up_ as well as the usual
>
> two facing "forward".
>
> Using just a little bit of digital logic (rather similar to what I was
> talking about in that other huge note way back when) Jon was able to come
> up with a behavior that is very nearly a solution to Bruce's challenge.
> If
> the top sensor "saw" light then the photovore worked just as you would
> expect, moving toward light. On the other hand, if the top sensor
> 'suddenly' was in the dark a new behavior would take over (subsuming the
> light chasing behavior, by the way). The photovore would do its best to
> go
> ANYWHERE quickly, but the ANYWHERE depended on the front sensors -- it
> would do its best to avoid the dark, "presuming" that darkness meant an
> obstacle.
>
> As a result, when the Deadly Darkness approached from above, the robot
> might zoom straight forward, spin left or right in place until it found a
> "good" place to run and then scoot or even simply zoom straight backwards.
>
> So... with the addition of a single sensor and a teensy bit of digital
> logic the robot was enormously more "adept" than a simple photovore.
>
>
>
> Zoz



13601 Mon, 17 Apr 2000 17:40:55 -0700 RE: REVERSER, Again! "'beam@sgiblab.sgi.com'" Wilf Rigter Aha! the reverser can't drive a 74ALS245 input when in the non-inverting
mode (through the 47K resistor). Try changing the 47K resistor to between a
resistor between 1K and 5K, but use the highest possible value.

wilf

> -----Original Message-----
> From: BUDSCOTT@aol.com [SMTP:BUDSCOTT@aol.com]
> Sent: Monday, April 17, 2000 4:28 PM
> To: beam@sgiblab.sgi.com
> Subject: REVERSER, Again!
>
> I've got my two bicores hooked up so that there is a reverser that will
> reverser just the front motor, the reverser is hooked up to a ALS245 motor
>
> driver, and the reverser is the 240 chip with the 47k resistors across the
>
> inverters, BUT for some reason, the front motor will only run when the
> thing
> is reversed! What's up with that?
>
> -Spencer
>
> <http://www.botic.com/users/beamstop>
>
> not a robot scientist
> not a college major
> not a grad student
> not a professor
> not a very organized person
> just Spencer (isn't that impressive enough?)



13602 Mon, 17 Apr 2000 18:22:01 -0700 [alt-beam] RE: You guys disappoint me (rules) BEAM David Simmons

Don Papp wrote:
>
> On Sun, 16 Apr 2000, Thomas Pilgaard wrote:

> > I will have to disagree. Making rules tends to limit peoples imagination and
> > leaves no room for imagination and innovation. Besides, different ways of
>
> Poppycock! Nothing stimulates imagination and innovation like a
> set of rules hung on the wall! :)

Yes and no! It is like writing poetry, use the forms (rules) to gain a
controlled understanding, then just break them all and see what happens.
Of course as I say this I am seeing someone opening a TV to see how it
works...

Some rules are useful, others just hinder growth.

Regards,
Dave


Home