Alt-BEAM Archive
Message #05791
To: beam@sgiblab.sgi.com
From: Ripter01@aol.com
Date: Thu, 19 Aug 1999 01:08:54 EDT
Subject: [alt-beam] Re: Edge detectors
The First robot I build, the WAO2 (From the Robot Store) Has what your
talking about build in, without the weights, What it dose is the whiskers are
bent to that when the robot in on the table it will close the switch but if
one of the whiskers goes over the edges of the table the switch opens and the
robot backs up/Turns whatever as not to fall off.
That me $2.50
-Ripter
In a message dated 8/18/99 4:39:41 PM US Eastern Standard Time,
arno.jansen@superconsult.nl writes:
> Just an idea I got while trying to sleep....(know what I mean???)
>
> Okay, now... I know there are bots with obstacle avoidance/detection. So,
> they will most likely not run into a wall. But, when you put them on a
> table, they might fall of the edge.......probably dead.
>
> But! (here's my little brainwave). It must be possible to make some
> edge-detectors. I haven't drawn a scheme or something, but this is what I
> think: It can some sort of leg. For example left front and one on the front
> right side. (ascii-1)
>
>
> \ /
> \ / <- edge detectors
> \ /
> --------------
> | bot |
> ----------------
> ascii-1 (I hope this will be well viewable...)
>
> Those legs must be stiff (not being able to bend) have some weight at their
> ends (those you remove from the namiki's). When the "edge-detector"
reaches
> an edge, the little weight will make it drop off, so it makes contact (just
> like a tactile sensor) and the bot can react and probably move backwards.
>
> For walkers you might want to use a pressure sensor, they're not expensive
> and they're small. So if you somehow manage to know when a walkers' leg is
> in the air, and when it's on the surface, you can "predict" what the
> difference of the value of the pressure sensor is, so whenthe leg is on the
> surface and the sensor's value is equal to the value it gives when the leg
> is in the air, you know, it's probably near a gap, or an edge and it could
> react and move in another direction.
>
> I am not (yet) able to try this, 'cause I'm still waiting for parts to
build
> my first photovore... and I haven't much spare parts yet. But if anyone
> starts thinking about this concept, this mail wasn't for nothing.
>
> So.. comments?
>
> Arno
>
> P.S. Or was I stupid to think I was the only one thinking of this
problem???
> :)
5792 Wed, 18 Aug 1999 22:50:47 -0700 [alt-beam] Re: Edge detectors beam@sgiblab.sgi.com Bruce Robinson Arno Jansen wrote:
> ...
> For walkers you might want to use a pressure sensor,
> they're not expensive and they're small. So if you
> somehow manage to know when a walkers' leg is in the
> air, and when it's on the surface, you can "predict"
> what the difference of the value of the pressure
> sensor is, so when the leg is on the surface and the
> sensor's value is equal to the value it gives when the
> leg is in the air, you know, it's probably near a gap,
> or an edge and it could react and move in another direction.
So, as I understand this, the pressure sensor tells you when the leg is
touching a surface. You predict when the leg OUGHT to be touching a
surface (perhaps by timing, or by a leg-position sensor), and if the
pressure sensors says the leg is NOT touching, you know you're about to
step in a hole, or off a cliff.
Not a bad idea at all. The pressure sensor can be a foot, instead of an
extra contact that can get caught on stuff. I don't recall seeing quite
that method on the net (but there is a lot out there I haven't found
yet).
> P.S. Or was I stupid to think I was the only
> one thinking of this problem???
Nope.
Regards,
Bruce
5793 Thu, 19 Aug 1999 08:19:48 +0200 [alt-beam] Re: Edge detectors "'beam@sgiblab.sgi.com'" "Van Zoelen, Bram AA SSI-TSEA-352"
> ----------
> From:
> arno.jansen@superconsult.nl[SMTP:arno.jansen@superconsult.nl]
> Reply To: beam@sgiblab.sgi.com
> Sent: Wednesday, August 18, 1999 11:33 PM
> To: beam@sgiblab.sgi.com
> Subject: Edge detectors
>
> Just an idea I got while trying to sleep....(know what I mean???)
>
Tell me, did you actually got some sleep that evening?
> Okay, now... I know there are bots with obstacle avoidance/detection. So,
> they will most likely not run into a wall. But, when you put them on a
> table, they might fall of the edge.......probably dead.
>
> But! (here's my little brainwave). It must be possible to make some
> edge-detectors. I haven't drawn a scheme or something, but this is what I
> think: It can some sort of leg. For example left front and one on the
> front
> right side. (ascii-1)
>
>
> \ /
> \ / <- edge detectors
> \ /
> --------------
> | bot |
> ----------------
> ascii-1 (I hope this will be well viewable...)
>
> Those legs must be stiff (not being able to bend) have some weight at
> their
> ends (those you remove from the namiki's). When the "edge-detector"
> reaches
> an edge, the little weight will make it drop off, so it makes contact
> (just
> like a tactile sensor) and the bot can react and probably move backwards.
>
Have you ever seen a two-motor bicore walker walk? It's a very bumpy ride.
The sensors needs a large free-movement space before it is allowed to
trigger. Even more, the gait of this walker is based on falling or
un-balance. So how will you detected the difference between the two. Also
the "dropped" sensor can get easily stuck if the surface isn't flat.
> For walkers you might want to use a pressure sensor, they're not expensive
> and they're small. So if you somehow manage to know when a walkers' leg is
> in the air, and when it's on the surface, you can "predict" what the
> difference of the value of the pressure sensor is, so whenthe leg is on
> the
> surface and the sensor's value is equal to the value it gives when the leg
> is in the air, you know, it's probably near a gap, or an edge and it could
> react and move in another direction.
>
A simple way to deal with the "find a firm foothole location" problem is to
put switches in the leg at the place where the foot is located. If the
switch is pressed then the foot has found a surface. The condition of this
footprint is an other matter of concern.
> I am not (yet) able to try this, 'cause I'm still waiting for parts to
> build
> my first photovore... and I haven't much spare parts yet. But if anyone
> starts thinking about this concept, this mail wasn't for nothing.
>
Oh, well. My answers wheren't well thought too.
> So.. comments?
>
> Arno
>
> P.S. Or was I stupid to think I was the only one thinking of this
> problem???
> :)
>
No. You can have a look in the archives to see if i am right :0)
Bram
Home