sonar pinging echo ideas
Archiver
Posts: 46,084
Hi,
..again typing aloud... Maybe a more elaborate model
is needed to process the signals?
Thinking of impact avoidance by your big robot a fast
echo must be a nearby hazard and so that signal needs
action now - so a fast ping rate needed. But does the
rate need to be constant? If no nearby object is
detected at a given time then a longer time can be
left between pings. First echo is always going to be
the nearest object, second says more about the larger
environment. Even if the transducer works backwards
(ie will send and recieve from its closed side) you
can detect retreat from a nearby wall by distance
increasing as the robot moves. This means that some
sort of closing velocity would be the thing to extract
rather than planin distance. The velocity would then
control both ping rate and action to take - would
progress quite nicely into a robot that runs away from
you!
...the 'extra' echos are real information and maybe
shouldn't be discarded so readily...
The echo could be detected as it will reveal a
distance related to the 'real' distance indicated by
the opposing transducer. This might be coded by a
'circular' lookup table that mimics the width of the
room. This could be determined on power-up and
obviously you might work this into a 3D model! The
robot would then be placed into this table. The
distance indicated by an echo should then 'run around'
the table in a sort of 'modulo' arithmetic fashion and
serve to confirm that the echo is indicative of so
many echo quanta.
If you can look at amplitude of return signal that
will carry some of that info too.
Adrian
=====
-
*********************************************
________________________________________________________________________
Want to chat instantly with your online friends? Get the FREE Yahoo!
Messenger http://uk.messenger.yahoo.com/
..again typing aloud... Maybe a more elaborate model
is needed to process the signals?
Thinking of impact avoidance by your big robot a fast
echo must be a nearby hazard and so that signal needs
action now - so a fast ping rate needed. But does the
rate need to be constant? If no nearby object is
detected at a given time then a longer time can be
left between pings. First echo is always going to be
the nearest object, second says more about the larger
environment. Even if the transducer works backwards
(ie will send and recieve from its closed side) you
can detect retreat from a nearby wall by distance
increasing as the robot moves. This means that some
sort of closing velocity would be the thing to extract
rather than planin distance. The velocity would then
control both ping rate and action to take - would
progress quite nicely into a robot that runs away from
you!
...the 'extra' echos are real information and maybe
shouldn't be discarded so readily...
The echo could be detected as it will reveal a
distance related to the 'real' distance indicated by
the opposing transducer. This might be coded by a
'circular' lookup table that mimics the width of the
room. This could be determined on power-up and
obviously you might work this into a 3D model! The
robot would then be placed into this table. The
distance indicated by an echo should then 'run around'
the table in a sort of 'modulo' arithmetic fashion and
serve to confirm that the echo is indicative of so
many echo quanta.
If you can look at amplitude of return signal that
will carry some of that info too.
Adrian
=====
-
*********************************************
________________________________________________________________________
Want to chat instantly with your online friends? Get the FREE Yahoo!
Messenger http://uk.messenger.yahoo.com/
Comments
(or guessed/gathered) information about the environment is a cool idea...
[noparse]:)[/noparse]
On another tangent, it occurred to me that maybe the problem could be
solved by reducing the number of sonar units from three to one and then
using a servo to pan the one unit across a 180 degree sweep and record at
intervals. The altering the resolution of the record intervals would
allow you to "paint a picture" of the obsticals in front of the bot with
varrying degrees of resolution. POssible reduce resolution as speed
increases and the, when you detect somthing, reduce speed and ramp up
resolution so you can carefully avoid it?
I've seen this "scanning sonar" concept used in a couple of designs. One
had a sonar module (like the polaroid I think) pointed straight up into a
sort of acoustic "perioscope" with its opening being swiveled from side
to side. Another design used small xmit and receive sensors side by side
on a small PC board (like the parallax unit described here:
http://www.parallax.com/downloads/documentation/nv/v3/col/nv84-sonic_sight_-_see\
ing_with_sound.pdf
This small PC board can then be mounted on a small "turret" attached to a
server and swept a programable distance. This is very versitile as the
speed of the sweep, the frequency (interval not acoustics) of the samples
and the area to be swept can all be software controlled.
This wouldn't solve the false echo problem so much as replace the process
that creates the false echo with one that might be able to provide you
with the same (or more) results... [noparse]:)[/noparse]
Vern
--
Vern Graner CNE/CNA/SSE | "If the network is down, then you're
Senior Systems Engineer | obviously incompetent so why are we
Texas Information Services | paying you? Of course, if the network
vern@t... www.txis.com | is up, then we obviously don't need
Cell 507-7851 Desk 328-8947 | you, so why are we paying you?" VLG
Adrian said:
> Hi,
>
> ..again typing aloud... Maybe a more elaborate model
> is needed to process the signals?
>
> Thinking of impact avoidance by your big robot a fast
> echo must be a nearby hazard and so that signal needs
> action now - so a fast ping rate needed. But does the
> rate need to be constant? If no nearby object is
> detected at a given time then a longer time can be
> left between pings. First echo is always going to be
> the nearest object, second says more about the larger
> environment. Even if the transducer works backwards
> (ie will send and recieve from its closed side) you
> can detect retreat from a nearby wall by distance
> increasing as the robot moves. This means that some
> sort of closing velocity would be the thing to extract
> rather than planin distance. The velocity would then
> control both ping rate and action to take - would
> progress quite nicely into a robot that runs away from
> you!
>
> ...the 'extra' echos are real information and maybe
> shouldn't be discarded so readily...
>
> The echo could be detected as it will reveal a
> distance related to the 'real' distance indicated by
> the opposing transducer. This might be coded by a
> 'circular' lookup table that mimics the width of the
> room. This could be determined on power-up and
> obviously you might work this into a 3D model! The
> robot would then be placed into this table. The
> distance indicated by an echo should then 'run around'
> the table in a sort of 'modulo' arithmetic fashion and
> serve to confirm that the echo is indicative of so
> many echo quanta.
>
> If you can look at amplitude of return signal that
> will carry some of that info too.
>
> Adrian
>
>
> =====
> -
> *********************************************
>
> ________________________________________________________________________
> Want to chat instantly with your online friends? Get the FREE Yahoo!
> Messenger http://uk.messenger.yahoo.com/
>
> To UNSUBSCRIBE, just send mail to:
> basicstamps-unsubscribe@yahoogroups.com
> from the same email address that you subscribed. Text in the Subject and
> Body of the message will be ignored.
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>
mounted on a disk which is rotated with a servo. The I/R units are
mounted 90 degrees from each other with the ultrasonic unit 45 degrees
between them. I can do obstacle avoidance or wall-following depending on
how I rotate the sensors.
On Mon, 2003-07-21 at 10:09, Vernon Graner wrote:
> Adrian: Interesting ideas, the use of the "false" echos to verify known
> (or guessed/gathered) information about the environment is a cool idea...
> [noparse]:)[/noparse]
>
>
> On another tangent, it occurred to me that maybe the problem could be
> solved by reducing the number of sonar units from three to one and then
> using a servo to pan the one unit across a 180 degree sweep and record at
> intervals. The altering the resolution of the record intervals would
> allow you to "paint a picture" of the obsticals in front of the bot with
> varrying degrees of resolution. POssible reduce resolution as speed
> increases and the, when you detect somthing, reduce speed and ramp up
> resolution so you can carefully avoid it?
>
> I've seen this "scanning sonar" concept used in a couple of designs. One
> had a sonar module (like the polaroid I think) pointed straight up into a
> sort of acoustic "perioscope" with its opening being swiveled from side
> to side. Another design used small xmit and receive sensors side by side
> on a small PC board (like the parallax unit described here:
>
>
http://www.parallax.com/downloads/documentation/nv/v3/col/nv84-sonic_sight_-_see\
ing_with_sound.pdf
>
> This small PC board can then be mounted on a small "turret" attached to a
> server and swept a programable distance. This is very versitile as the
> speed of the sweep, the frequency (interval not acoustics) of the samples
> and the area to be swept can all be software controlled.
>
> This wouldn't solve the false echo problem so much as replace the process
> that creates the false echo with one that might be able to provide you
> with the same (or more) results... [noparse]:)[/noparse]
>
> Vern