Loading ...
Sorry, an error occurred while loading the content.

Advanced Beam Head Technology Question

Expand Messages
  • S. G.
    Hi all, Greetings from Texas! I know a lot of you (including myself) have built powersmart beam heads, or maybe just simple bicore beam heads. I have enjoyed
    Message 1 of 14 , Nov 29, 2000
    • 0 Attachment
      Hi all,
      Greetings from Texas!

      I know a lot of you (including myself) have built powersmart beam heads, or
      maybe just simple bicore beam heads. I have enjoyed this technology,
      (thanks Wilf!) and have even sought (with Bruce's help) to incorporate one
      into a two motor walker. It's a work still in progress. ( ok ok, its on the
      shelf now, but I'll get there soon, I promise!)

      Now bear with me here, and please keep the flames regarding my stupidy on
      this matter to a minimum. :)
      Is it possible to do more than have the head turn towards a brighter source
      of light?
      The case I am thinking of is a head that turns towards any light change in
      the room. The theory of operation is that if something moves, then the
      light characteristics are going to change in that area. The bigger the
      change, the more apt the head is to look at it.
      Now, with a microprocessor the HEAD can determine when to move by can useing
      the LDRs as sort of a sensor which could generate a range of values based on
      the light level.
      This has been done with great results, as seen at:
      http://www.nwlink.com/~kevinro/lowresv.html
      which I have mentioned here before. BUT, I got to thinking, Why not use
      more LDRs in a vision configuration (such as in the URL mentioned above) and
      use lenses in the same way to bring light into focus, and the determine the
      change of light in the room in much the same way, but by using an analog
      (BEAM) head circuit?
      Seems to be the concept is close to the same, the encreased LDRs simply
      amplify the sensitivity, when they are brought into focus.
      The mechanics (configuration?) has already been done. Seems like a lot of
      work has already been accomplished for us. (again I'm referring to the above
      URL) What sort of modified power-smart head circuit would work on such a
      project?
      OR,
      am I simply way off base here? Seems to me that such a beast is possible.
      I am willing to invest atever $$ it takes to see if such a thing can happen,
      As I have some ideas for a bot like this.
      Comments?
      -Sparky
    • Daniel Grace
      Interesting idea, and definately would appear biological! I would at least try to go with the light sensor with the quickest responce time (I don t know if
      Message 2 of 14 , Nov 29, 2000
      • 0 Attachment
        Interesting idea, and definately would appear
        biological! I would at least try to go with the light
        sensor with the quickest responce time (I don't know
        if that's LDRs, Solar Cells, or maybe photodiodes...),
        amplify the signal recieved with capacitively coupled
        amplifiers, throw another capacitor in there
        somewhere, and use a Nv and Nu to detect the change.
        The first half of the BeamSonic circuit does exactly
        this to detect sounds.

        This may not work with a light sensor, they don't
        usually respond really quickly, and even if they do,
        there may be a better way to do it.

        Anyone else?

        ~Daniel

        --- "S. G." <sgarvin@...> wrote:
        > Hi all,
        > Greetings from Texas!
        >
        > I know a lot of you (including myself) have built
        > powersmart beam heads, or
        > maybe just simple bicore beam heads. I have enjoyed
        > this technology,
        > (thanks Wilf!) and have even sought (with Bruce's
        > help) to incorporate one
        > into a two motor walker. It's a work still in
        > progress. ( ok ok, its on the
        > shelf now, but I'll get there soon, I promise!)
        >
        > Now bear with me here, and please keep the flames
        > regarding my stupidy on
        > this matter to a minimum. :)
        > Is it possible to do more than have the head turn
        > towards a brighter source
        > of light?
        > The case I am thinking of is a head that turns
        > towards any light change in
        > the room. The theory of operation is that if
        > something moves, then the
        > light characteristics are going to change in that
        > area. The bigger the
        > change, the more apt the head is to look at it.
        > Now, with a microprocessor the HEAD can determine
        > when to move by can useing
        > the LDRs as sort of a sensor which could generate a
        > range of values based on
        > the light level.
        > This has been done with great results, as seen at:
        > http://www.nwlink.com/~kevinro/lowresv.html
        > which I have mentioned here before. BUT, I got to
        > thinking, Why not use
        > more LDRs in a vision configuration (such as in the
        > URL mentioned above) and
        > use lenses in the same way to bring light into
        > focus, and the determine the
        > change of light in the room in much the same way,
        > but by using an analog
        > (BEAM) head circuit?
        > Seems to be the concept is close to the same, the
        > encreased LDRs simply
        > amplify the sensitivity, when they are brought into
        > focus.
        > The mechanics (configuration?) has already been
        > done. Seems like a lot of
        > work has already been accomplished for us. (again
        > I'm referring to the above
        > URL) What sort of modified power-smart head
        > circuit would work on such a
        > project?
        > OR,
        > am I simply way off base here? Seems to me that
        > such a beast is possible.
        > I am willing to invest atever $$ it takes to see if
        > such a thing can happen,
        > As I have some ideas for a bot like this.
        > Comments?
        > -Sparky
        >
        >
        >
        >
        > -------------------------- eGroups Sponsor
        >
        > To unsubscribe from this group, send an email to:
        > beam-unsubscribe@egroups.com
        >
        >
        >


        =====
        ICQ # 39402143
        http://www.negativeview.com
        [non-beam]

        __________________________________________________
        Do You Yahoo!?
        Yahoo! Shopping - Thousands of Stores. Millions of Products.
        http://shopping.yahoo.com/
      • John Simmons
        Hi Y all, Greetings from Texas! (We re everywhere we re everywhere, dunt da da dunt DUNT!) S. G. that sounds very intriguing. What would be the reference to
        Message 3 of 14 , Nov 29, 2000
        • 0 Attachment
          Hi Y'all,
          Greetings from Texas! (We're everywhere we're everywhere, dunt da da dunt
          DUNT!)
          S. G. that sounds very intriguing. What would be the reference to compare
          the new light level too in order to calculate the change? Or what would
          contain the reference data?
          1) You could digitally store a number which represents a light condition
          then compare the new one to that then decode it for some action.
          2) You could store it in Wilf's memory chain for a period of time then
          act before it times out and loses that info.
          3)You could have a simple reference voltage it compares to.

          4)What would the head be looking for, or what would be the action that
          results? Would this be a "smarter" head? Maybe it could control more
          functions.
          It could a)Direct the bot's travel direction
          b)Direct a sensor array toward some desired light quality
          c)Keep a cell array pointed toward the best light source
          d)It could keep cycling through each function continually
          adjusting each part.
          e)It could aim a weapons platform for use in a robot wars
          game.
          f)It could move a platform just to look cool like an
          insect looking around.
          What else.....? John S.


          >From: "S. G." <sgarvin@...>
          >Reply-To: beam@egroups.com
          >To: <beam@egroups.com>
          >Subject: [beam] Advanced Beam Head Technology Question
          >Date: Wed, 29 Nov 2000 22:28:41 -0600
          >
          >Hi all,
          >Greetings from Texas!
          >
          >I know a lot of you (including myself) have built powersmart beam heads,
          >or
          >maybe just simple bicore beam heads. I have enjoyed this technology,
          >(thanks Wilf!) and have even sought (with Bruce's help) to incorporate one
          >into a two motor walker. It's a work still in progress. ( ok ok, its on the
          >shelf now, but I'll get there soon, I promise!)
          >
          >Now bear with me here, and please keep the flames regarding my stupidy on
          >this matter to a minimum. :)
          >Is it possible to do more than have the head turn towards a brighter source
          >of light?
          >The case I am thinking of is a head that turns towards any light change in
          >the room. The theory of operation is that if something moves, then the
          >light characteristics are going to change in that area. The bigger the
          >change, the more apt the head is to look at it.
          >Now, with a microprocessor the HEAD can determine when to move by can
          >useing
          >the LDRs as sort of a sensor which could generate a range of values based
          >on
          >the light level.
          >This has been done with great results, as seen at:
          >http://www.nwlink.com/~kevinro/lowresv.html
          >which I have mentioned here before. BUT, I got to thinking, Why not use
          >more LDRs in a vision configuration (such as in the URL mentioned above)
          >and
          >use lenses in the same way to bring light into focus, and the determine the
          >change of light in the room in much the same way, but by using an analog
          >(BEAM) head circuit?
          >Seems to be the concept is close to the same, the encreased LDRs simply
          >amplify the sensitivity, when they are brought into focus.
          >The mechanics (configuration?) has already been done. Seems like a lot of
          >work has already been accomplished for us. (again I'm referring to the
          >above
          >URL) What sort of modified power-smart head circuit would work on such a
          >project?
          >OR,
          >am I simply way off base here? Seems to me that such a beast is possible.
          >I am willing to invest atever $$ it takes to see if such a thing can
          >happen,
          >As I have some ideas for a bot like this.
          >Comments?
          >-Sparky
          >
          >
          >

          _____________________________________________________________________________________
          Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
        • John Simmons
          Oooh! OOoh! What if the head was comprised of like three layers of bands? The middle ring or band looks for light, the top band scans the IR region, the lower
          Message 4 of 14 , Nov 29, 2000
          • 0 Attachment
            Oooh! OOoh! What if the head was comprised of like three layers of bands?
            The middle ring or band looks for light, the top band scans the IR region,
            the lower band scans or turns toward noise levels, radio waves, gravity
            waves! All these bands rotating around independently with different sensor
            shapes would look way cool.
            It's like you took a basketball and cut it into three or four layers.
            Overall it is still ball shaped but the diff. layers are scanning like
            lizard eyes moving independently. Too neat-o! John S.


            >From: Daniel Grace <lifebytes_98@...>
            >Reply-To: beam@egroups.com
            >To: beam@egroups.com
            >Subject: Re: [beam] Advanced Beam Head Technology Question
            >Date: Wed, 29 Nov 2000 20:49:56 -0800 (PST)
            >
            >Interesting idea, and definately would appear
            >biological! I would at least try to go with the light
            >sensor with the quickest responce time (I don't know
            >if that's LDRs, Solar Cells, or maybe photodiodes...),
            >amplify the signal recieved with capacitively coupled
            >amplifiers, throw another capacitor in there
            >somewhere, and use a Nv and Nu to detect the change.
            >The first half of the BeamSonic circuit does exactly
            >this to detect sounds.
            >
            >This may not work with a light sensor, they don't
            >usually respond really quickly, and even if they do,
            >there may be a better way to do it.
            >
            >Anyone else?
            >
            >~Daniel
            >
            >--- "S. G." <sgarvin@...> wrote:
            > > Hi all,
            > > Greetings from Texas!
            > >
            > > I know a lot of you (including myself) have built
            > > powersmart beam heads, or
            > > maybe just simple bicore beam heads. I have enjoyed
            > > this technology,
            > > (thanks Wilf!) and have even sought (with Bruce's
            > > help) to incorporate one
            > > into a two motor walker. It's a work still in
            > > progress. ( ok ok, its on the
            > > shelf now, but I'll get there soon, I promise!)
            > >
            > > Now bear with me here, and please keep the flames
            > > regarding my stupidy on
            > > this matter to a minimum. :)
            > > Is it possible to do more than have the head turn
            > > towards a brighter source
            > > of light?
            > > The case I am thinking of is a head that turns
            > > towards any light change in
            > > the room. The theory of operation is that if
            > > something moves, then the
            > > light characteristics are going to change in that
            > > area. The bigger the
            > > change, the more apt the head is to look at it.
            > > Now, with a microprocessor the HEAD can determine
            > > when to move by can useing
            > > the LDRs as sort of a sensor which could generate a
            > > range of values based on
            > > the light level.
            > > This has been done with great results, as seen at:
            > > http://www.nwlink.com/~kevinro/lowresv.html
            > > which I have mentioned here before. BUT, I got to
            > > thinking, Why not use
            > > more LDRs in a vision configuration (such as in the
            > > URL mentioned above) and
            > > use lenses in the same way to bring light into
            > > focus, and the determine the
            > > change of light in the room in much the same way,
            > > but by using an analog
            > > (BEAM) head circuit?
            > > Seems to be the concept is close to the same, the
            > > encreased LDRs simply
            > > amplify the sensitivity, when they are brought into
            > > focus.
            > > The mechanics (configuration?) has already been
            > > done. Seems like a lot of
            > > work has already been accomplished for us. (again
            > > I'm referring to the above
            > > URL) What sort of modified power-smart head
            > > circuit would work on such a
            > > project?
            > > OR,
            > > am I simply way off base here? Seems to me that
            > > such a beast is possible.
            > > I am willing to invest atever $$ it takes to see if
            > > such a thing can happen,
            > > As I have some ideas for a bot like this.
            > > Comments?
            > > -Sparky
            > >
            > >
            > >
            > >
            > > -------------------------- eGroups Sponsor
            > >
            > > To unsubscribe from this group, send an email to:
            > > beam-unsubscribe@egroups.com
            > >
            > >
            > >
            >
            >
            >=====
            >ICQ # 39402143
            >http://www.negativeview.com
            >[non-beam]
            >
            >__________________________________________________
            >Do You Yahoo!?
            >Yahoo! Shopping - Thousands of Stores. Millions of Products.
            >http://shopping.yahoo.com/

            _____________________________________________________________________________________
            Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
          • S. G.
            ... YipeeYI YAA! ... results? actually all that you have mentioned and more, but for starters, I gotta go ... around. you gotta admit a mechanical beastie
            Message 5 of 14 , Nov 29, 2000
            • 0 Attachment
              > Greetings from Texas! (We're everywhere we're everywhere, dunt da da dunt
              > DUNT!)

              YipeeYI YAA!

              >What would be the reference to compare
              > the new light level too in order to calculate the change?

              I was thinking along the lines of this one, number 3:

              > 3)You could have a simple reference voltage it >compares to.

              >What would the head be looking for, or what would be >the action that
              results?

              actually all that you have mentioned and more, but for starters, I gotta go
              with the one you labled "f" :

              >f)It could move a platform just to look cool like an >insect looking
              around.

              you gotta admit a mechanical beastie that raises its head and looks as you
              as you come into a room and who turns to watch you move about has gotta be
              cool.
              -Sparky
            • Wilf Rigter
              Interesting idea: the LRVS head action depends on its software algorithm. Unlike a beam head which is continuously tracking, scanning and correcting the error
              Message 6 of 14 , Nov 29, 2000
              • 0 Attachment
                Interesting idea: the LRVS head action depends on its software algorithm.
                Unlike a beam head which is continuously tracking, scanning and correcting
                the error or difference signal between its two sensors, a LRVS bot uses a
                software algorithm that scans the sensor array and makes a decision based on
                memorized data to change its orientation.
                The software for tracking the brightest light is probably the simplest and
                can be virtually executed in real time by sending a correction signal to the
                servos to center the brightest light in the field of view to the middle of
                the sensor array. This is very similar to the light controlled analog servo
                loop of BEAM heads. If the LRVS head is tracking a particular light, it can
                compare it with other light sources that enter it's field of vision and
                switch to a brighter light source. With software, the LRVS head can spin
                around to survey 360 degrees and decide which of several lights (in front or
                behind) is the brightest and then steer the bot (or head) towards it. By
                scanning or rotating the sensor array independent of the head or moving
                platform, it can take a survey and then act on that stored information. So
                once per sensor revolution or scan, the brightest source is identified and
                the head or platform turned towards it. Since this is no longer done in real
                time, the brain makes decisions based on a memory map of the world.

                To detect "movement" or "change", something else is needed: the array must
                be held steady with respect to the world or else, the way the array sees it,
                when the sensors are moving, the whole world would appear to move or change
                with respect to it. A histogram of variations in light levels on each sensor
                are accumulated and the servos are directed to adjust the center of view to
                the sensor with the largest changes (eg flashing LED)until the largest
                change occurs on the center sensor. However, every time it changes position
                it has to dump the sensor history, hold still and acquire new "change" data.
                In addition, for obvious reasons the sensors have to be "blind to change"
                while the sensors are being moved. A really smart head could do the matrix
                calculations to transfer the history of sensor data that remains in the
                field of view.

                Anyway you get the picture. For all but the simplest bright light tracking
                case, this software / memory map can not be easily adapted to the analog
                servo loop circuits of Beam heads.

                A different approach would use an array and a neural/nervous network to
                weight the outputs of the sensors and generate a 2 phase error signal to
                drive the motor servo loop. By AC coupling and rectifying the sensor outputs
                (or equivalent) changes can be detected and "compared" to the changes in
                other sensors. Then a pulse is generated with a phase and duration
                proportional to angle and proximity of the sensor of the largest change to
                the sensor arry center. The effect of that will be to calculate the error
                signal and execute the correction after which the process repeats. During
                the execution phase that sensor "change" data is blocked. So the servo loop
                process is no longer continuous but divided into discrete time steps.

                Next a few circuits that use a sensor array to turn beam head towards the
                brightests light in the field of view.

                wilf

                -----Original Message-----
                From: S. G.
                To: beam@egroups.com
                Sent: 11/29/00 8:28 PM
                Subject: [beam] Advanced Beam Head Technology Question

                Hi all,
                Greetings from Texas!

                I know a lot of you (including myself) have built powersmart beam heads,
                or
                maybe just simple bicore beam heads. I have enjoyed this technology,
                (thanks Wilf!) and have even sought (with Bruce's help) to incorporate
                one
                into a two motor walker. It's a work still in progress. ( ok ok, its on
                the
                shelf now, but I'll get there soon, I promise!)

                Now bear with me here, and please keep the flames regarding my stupidy
                on
                this matter to a minimum. :)
                Is it possible to do more than have the head turn towards a brighter
                source
                of light?
                The case I am thinking of is a head that turns towards any light change
                in
                the room. The theory of operation is that if something moves, then the
                light characteristics are going to change in that area. The bigger the
                change, the more apt the head is to look at it.
                Now, with a microprocessor the HEAD can determine when to move by can
                useing
                the LDRs as sort of a sensor which could generate a range of values
                based on
                the light level.
                This has been done with great results, as seen at:
                http://www.nwlink.com/~kevinro/lowresv.html
                which I have mentioned here before. BUT, I got to thinking, Why not
                use
                more LDRs in a vision configuration (such as in the URL mentioned above)
                and
                use lenses in the same way to bring light into focus, and the determine
                the
                change of light in the room in much the same way, but by using an analog
                (BEAM) head circuit?
                Seems to be the concept is close to the same, the encreased LDRs simply
                amplify the sensitivity, when they are brought into focus.
                The mechanics (configuration?) has already been done. Seems like a lot
                of
                work has already been accomplished for us. (again I'm referring to the
                above
                URL) What sort of modified power-smart head circuit would work on such
                a
                project?
                OR,
                am I simply way off base here? Seems to me that such a beast is
                possible.
                I am willing to invest atever $$ it takes to see if such a thing can
                happen,
                As I have some ideas for a bot like this.
                Comments?
                -Sparky
              • William Cox
                A phototransistor responds quickly and has a linear resistance based on light. -William ... From: Daniel Grace To:
                Message 7 of 14 , Nov 30, 2000
                • 0 Attachment
                  A phototransistor responds quickly and has a linear resistance based on
                  light.
                  -William

                  ----- Original Message -----
                  From: Daniel Grace <lifebytes_98@...>
                  To: <beam@egroups.com>
                  Sent: Wednesday, November 29, 2000 11:49 PM
                  Subject: Re: [beam] Advanced Beam Head Technology Question


                  > Interesting idea, and definately would appear
                  > biological! I would at least try to go with the light
                  > sensor with the quickest responce time (I don't know
                  > if that's LDRs, Solar Cells, or maybe photodiodes...),
                  > amplify the signal recieved with capacitively coupled
                  > amplifiers, throw another capacitor in there
                  > somewhere, and use a Nv and Nu to detect the change.
                  > The first half of the BeamSonic circuit does exactly
                  > this to detect sounds.
                  >
                  > This may not work with a light sensor, they don't
                  > usually respond really quickly, and even if they do,
                  > there may be a better way to do it.
                  >
                  > Anyone else?
                  >
                  > ~Daniel
                  >
                  > --- "S. G." <sgarvin@...> wrote:
                  > > Hi all,
                  > > Greetings from Texas!
                  > >
                  > > I know a lot of you (including myself) have built
                  > > powersmart beam heads, or
                  > > maybe just simple bicore beam heads. I have enjoyed
                  > > this technology,
                  > > (thanks Wilf!) and have even sought (with Bruce's
                  > > help) to incorporate one
                  > > into a two motor walker. It's a work still in
                  > > progress. ( ok ok, its on the
                  > > shelf now, but I'll get there soon, I promise!)
                  > >
                  > > Now bear with me here, and please keep the flames
                  > > regarding my stupidy on
                  > > this matter to a minimum. :)
                  > > Is it possible to do more than have the head turn
                  > > towards a brighter source
                  > > of light?
                  > > The case I am thinking of is a head that turns
                  > > towards any light change in
                  > > the room. The theory of operation is that if
                  > > something moves, then the
                  > > light characteristics are going to change in that
                  > > area. The bigger the
                  > > change, the more apt the head is to look at it.
                  > > Now, with a microprocessor the HEAD can determine
                  > > when to move by can useing
                  > > the LDRs as sort of a sensor which could generate a
                  > > range of values based on
                  > > the light level.
                  > > This has been done with great results, as seen at:
                  > > http://www.nwlink.com/~kevinro/lowresv.html
                  > > which I have mentioned here before. BUT, I got to
                  > > thinking, Why not use
                  > > more LDRs in a vision configuration (such as in the
                  > > URL mentioned above) and
                  > > use lenses in the same way to bring light into
                  > > focus, and the determine the
                  > > change of light in the room in much the same way,
                  > > but by using an analog
                  > > (BEAM) head circuit?
                  > > Seems to be the concept is close to the same, the
                  > > encreased LDRs simply
                  > > amplify the sensitivity, when they are brought into
                  > > focus.
                  > > The mechanics (configuration?) has already been
                  > > done. Seems like a lot of
                  > > work has already been accomplished for us. (again
                  > > I'm referring to the above
                  > > URL) What sort of modified power-smart head
                  > > circuit would work on such a
                  > > project?
                  > > OR,
                  > > am I simply way off base here? Seems to me that
                  > > such a beast is possible.
                  > > I am willing to invest atever $$ it takes to see if
                  > > such a thing can happen,
                  > > As I have some ideas for a bot like this.
                  > > Comments?
                  > > -Sparky
                  > >
                  > >
                  > >
                  > >
                  > > -------------------------- eGroups Sponsor
                  > >
                  > > To unsubscribe from this group, send an email to:
                  > > beam-unsubscribe@egroups.com
                  > >
                  > >
                  > >
                  >
                  >
                  > =====
                  > ICQ # 39402143
                  > http://www.negativeview.com
                  > [non-beam]
                  >
                  > __________________________________________________
                  > Do You Yahoo!?
                  > Yahoo! Shopping - Thousands of Stores. Millions of Products.
                  > http://shopping.yahoo.com/
                  >
                  >
                  > To unsubscribe from this group, send an email to:
                  > beam-unsubscribe@egroups.com
                  >
                  >
                  >
                  >
                • William Cox
                  Instead of all this, couldn t you just have a normal head, that instead of comparing the two light sensors output in a comparitor fashion, would look for a
                  Message 8 of 14 , Nov 30, 2000
                  • 0 Attachment
                    Instead of all this, couldn't you just have a normal head, that instead of
                    comparing the two light sensors output in a comparitor fashion, would look
                    for a spike of light? A gradual change of values wouldn't affect the
                    positions of the head, only a rapid change of light levels. This rapid
                    change wouldn't necessarily have to be over a large range of values, so long
                    as it was rapid. The head would be turned in the direction of the 'eye' that
                    saw the largest/fastest spike/change in light levels.
                    As someone commented, the BEAM Sonic circuit acts in a similar way. It
                    doesn't necessarily track sound, but a change in sound (I think). A hand
                    clap would do more to affect it than would a continuous hum. This idea would
                    be prefect for modification to a movement tracking head.
                    -William


                    ----- Original Message -----
                    From: Wilf Rigter <Wilf.Rigter@...>
                    To: 'S. G. ' <sgarvin@...>; <beam@egroups.com>
                    Sent: Thursday, November 30, 2000 1:26 AM
                    Subject: RE: [beam] Advanced Beam Head Technology Question


                    > Interesting idea: the LRVS head action depends on its software algorithm.
                    > Unlike a beam head which is continuously tracking, scanning and
                    correcting
                    > the error or difference signal between its two sensors, a LRVS bot uses a
                    > software algorithm that scans the sensor array and makes a decision based
                    on
                    > memorized data to change its orientation.
                    > The software for tracking the brightest light is probably the simplest and
                    > can be virtually executed in real time by sending a correction signal to
                    the
                    > servos to center the brightest light in the field of view to the middle of
                    > the sensor array. This is very similar to the light controlled analog
                    servo
                    > loop of BEAM heads. If the LRVS head is tracking a particular light, it
                    can
                    > compare it with other light sources that enter it's field of vision and
                    > switch to a brighter light source. With software, the LRVS head can spin
                    > around to survey 360 degrees and decide which of several lights (in front
                    or
                    > behind) is the brightest and then steer the bot (or head) towards it. By
                    > scanning or rotating the sensor array independent of the head or moving
                    > platform, it can take a survey and then act on that stored information. So
                    > once per sensor revolution or scan, the brightest source is identified and
                    > the head or platform turned towards it. Since this is no longer done in
                    real
                    > time, the brain makes decisions based on a memory map of the world.
                    >
                    > To detect "movement" or "change", something else is needed: the array must
                    > be held steady with respect to the world or else, the way the array sees
                    it,
                    > when the sensors are moving, the whole world would appear to move or
                    change
                    > with respect to it. A histogram of variations in light levels on each
                    sensor
                    > are accumulated and the servos are directed to adjust the center of view
                    to
                    > the sensor with the largest changes (eg flashing LED)until the largest
                    > change occurs on the center sensor. However, every time it changes
                    position
                    > it has to dump the sensor history, hold still and acquire new "change"
                    data.
                    > In addition, for obvious reasons the sensors have to be "blind to change"
                    > while the sensors are being moved. A really smart head could do the matrix
                    > calculations to transfer the history of sensor data that remains in the
                    > field of view.
                    >
                    > Anyway you get the picture. For all but the simplest bright light tracking
                    > case, this software / memory map can not be easily adapted to the analog
                    > servo loop circuits of Beam heads.
                    >
                    > A different approach would use an array and a neural/nervous network to
                    > weight the outputs of the sensors and generate a 2 phase error signal to
                    > drive the motor servo loop. By AC coupling and rectifying the sensor
                    outputs
                    > (or equivalent) changes can be detected and "compared" to the changes in
                    > other sensors. Then a pulse is generated with a phase and duration
                    > proportional to angle and proximity of the sensor of the largest change to
                    > the sensor arry center. The effect of that will be to calculate the error
                    > signal and execute the correction after which the process repeats. During
                    > the execution phase that sensor "change" data is blocked. So the servo
                    loop
                    > process is no longer continuous but divided into discrete time steps.
                    >
                    > Next a few circuits that use a sensor array to turn beam head towards the
                    > brightests light in the field of view.
                    >
                    > wilf
                    >
                    > -----Original Message-----
                    > From: S. G.
                    > To: beam@egroups.com
                    > Sent: 11/29/00 8:28 PM
                    > Subject: [beam] Advanced Beam Head Technology Question
                    >
                    > Hi all,
                    > Greetings from Texas!
                    >
                    > I know a lot of you (including myself) have built powersmart beam heads,
                    > or
                    > maybe just simple bicore beam heads. I have enjoyed this technology,
                    > (thanks Wilf!) and have even sought (with Bruce's help) to incorporate
                    > one
                    > into a two motor walker. It's a work still in progress. ( ok ok, its on
                    > the
                    > shelf now, but I'll get there soon, I promise!)
                    >
                    > Now bear with me here, and please keep the flames regarding my stupidy
                    > on
                    > this matter to a minimum. :)
                    > Is it possible to do more than have the head turn towards a brighter
                    > source
                    > of light?
                    > The case I am thinking of is a head that turns towards any light change
                    > in
                    > the room. The theory of operation is that if something moves, then the
                    > light characteristics are going to change in that area. The bigger the
                    > change, the more apt the head is to look at it.
                    > Now, with a microprocessor the HEAD can determine when to move by can
                    > useing
                    > the LDRs as sort of a sensor which could generate a range of values
                    > based on
                    > the light level.
                    > This has been done with great results, as seen at:
                    > http://www.nwlink.com/~kevinro/lowresv.html
                    > which I have mentioned here before. BUT, I got to thinking, Why not
                    > use
                    > more LDRs in a vision configuration (such as in the URL mentioned above)
                    > and
                    > use lenses in the same way to bring light into focus, and the determine
                    > the
                    > change of light in the room in much the same way, but by using an analog
                    > (BEAM) head circuit?
                    > Seems to be the concept is close to the same, the encreased LDRs simply
                    > amplify the sensitivity, when they are brought into focus.
                    > The mechanics (configuration?) has already been done. Seems like a lot
                    > of
                    > work has already been accomplished for us. (again I'm referring to the
                    > above
                    > URL) What sort of modified power-smart head circuit would work on such
                    > a
                    > project?
                    > OR,
                    > am I simply way off base here? Seems to me that such a beast is
                    > possible.
                    > I am willing to invest atever $$ it takes to see if such a thing can
                    > happen,
                    > As I have some ideas for a bot like this.
                    > Comments?
                    > -Sparky
                    >
                    >
                    >
                    > To unsubscribe from this group, send an email to:
                    > beam-unsubscribe@egroups.com
                    >
                    >
                    >
                    >
                  • S. G.
                    ... interesting concept! but i think it would be more interesting if the device took in a scan about 30 degrees vertical , 45 degrees horizonal (left and
                    Message 9 of 14 , Nov 30, 2000
                    • 0 Attachment
                      >All these bands rotating around independently with
                      >different sensor shapes would look way cool.

                      interesting concept!
                      but i think it would be more interesting if the device took in a scan about
                      30 degrees vertical , 45 degrees horizonal (left and right) and stayed
                      completly static in relation to its inviroment until something changed.
                      In other words, was compleatly still until someone walked into the room.
                      Much like a power smart head. It doesn't move until a greater source of
                      light hits one of the sensors.
                      This is also really great on battery consumption.
                      -Sparky
                    • S. G.
                      ... Bingo! The difference I was looking for , was using a LDR or Photo transistor, or Photo Diode array (eight + in number?) so that it could determine changes
                      Message 10 of 14 , Nov 30, 2000
                      • 0 Attachment
                        > Instead of all this, couldn't you just have a normal head,
                        >that instead of comparing the two light sensors output in
                        >a comparitor fashion, would look for a spike of light? A
                        > gradual change of values wouldn't affect the
                        > positions of the head, only a rapid change of light >levels.

                        Bingo!
                        The difference I was looking for , was using a LDR or Photo transistor, or
                        Photo Diode array (eight + in number?) so that it could determine changes
                        across the room. For example, my PSHead, using two LDRs can see a surface
                        with light on it from about 8 inches away, but thats about it. What we are
                        talking about is looking for a change in light levels that are measured in
                        feet from the device.
                        If there is only one light in the room it will turn towards it, no matter
                        how far it is (up to 12 feet?) but only if there is no ambient (sp) light in
                        the room.
                        If a voltage detector can detect when voltage gets to a certain level,
                        (1381) is there one that can detext when voltage drops? such as when light
                        dependant resistors increase resistance ? wait, thats only current.
                        oops.
                        -Sparky
                      • Wilf Rigter
                        Intead of all this , meaning the LRVS system? The body heat motion detectors inherently can only detect change and may be useful for this application. The
                        Message 11 of 14 , Nov 30, 2000
                        • 0 Attachment
                          "Intead of all this", meaning the LRVS system?

                          The body heat motion detectors inherently can only detect change and may be
                          useful for this application. The Beam Sonic circuit detects sounds of the
                          right amplitude, frequency and duration and generates one charge transfer
                          per sound pulse.
                          Several sound pulses must be counted before the Sonic circuit sends it's own
                          sound pulse (chirp). During the time it chirps the Sonic circuit is
                          resetting the "counter" and ignores sound sources including itself. A LDR or
                          PD may be substituted for the microphone and a LED for the piezo beeper to
                          make the Sonic circuit repond to and generate with light pulses.
                          Alternatively two sonic circuits can be used with a motor connected between
                          the outputs and which circuit gets more sound or light hits determines the
                          direction of motor rotation. Ergo the Sonic Head.

                          wilf


                          -----Original Message-----
                          From: William Cox [mailto:william@...]
                          Sent: Thursday, November 30, 2000 7:10 AM
                          To: beam@egroups.com
                          Subject: Re: [beam] Advanced Beam Head Technology Question


                          Instead of all this, couldn't you just have a normal head, that instead of
                          comparing the two light sensors output in a comparitor fashion, would look
                          for a spike of light? A gradual change of values wouldn't affect the
                          positions of the head, only a rapid change of light levels. This rapid
                          change wouldn't necessarily have to be over a large range of values, so long
                          as it was rapid. The head would be turned in the direction of the 'eye' that
                          saw the largest/fastest spike/change in light levels.
                          As someone commented, the BEAM Sonic circuit acts in a similar way. It
                          doesn't necessarily track sound, but a change in sound (I think). A hand
                          clap would do more to affect it than would a continuous hum. This idea would
                          be prefect for modification to a movement tracking head.
                          -William


                          ----- Original Message -----
                          From: Wilf Rigter <Wilf.Rigter@...>
                          To: 'S. G. ' <sgarvin@...>; <beam@egroups.com>
                          Sent: Thursday, November 30, 2000 1:26 AM
                          Subject: RE: [beam] Advanced Beam Head Technology Question


                          > Interesting idea: the LRVS head action depends on its software algorithm.
                          > Unlike a beam head which is continuously tracking, scanning and
                          correcting
                          > the error or difference signal between its two sensors, a LRVS bot uses a
                          > software algorithm that scans the sensor array and makes a decision based
                          on
                          > memorized data to change its orientation.
                          > The software for tracking the brightest light is probably the simplest and
                          > can be virtually executed in real time by sending a correction signal to
                          the
                          > servos to center the brightest light in the field of view to the middle of
                          > the sensor array. This is very similar to the light controlled analog
                          servo
                          > loop of BEAM heads. If the LRVS head is tracking a particular light, it
                          can
                          > compare it with other light sources that enter it's field of vision and
                          > switch to a brighter light source. With software, the LRVS head can spin
                          > around to survey 360 degrees and decide which of several lights (in front
                          or
                          > behind) is the brightest and then steer the bot (or head) towards it. By
                          > scanning or rotating the sensor array independent of the head or moving
                          > platform, it can take a survey and then act on that stored information. So
                          > once per sensor revolution or scan, the brightest source is identified and
                          > the head or platform turned towards it. Since this is no longer done in
                          real
                          > time, the brain makes decisions based on a memory map of the world.
                          >
                          > To detect "movement" or "change", something else is needed: the array must
                          > be held steady with respect to the world or else, the way the array sees
                          it,
                          > when the sensors are moving, the whole world would appear to move or
                          change
                          > with respect to it. A histogram of variations in light levels on each
                          sensor
                          > are accumulated and the servos are directed to adjust the center of view
                          to
                          > the sensor with the largest changes (eg flashing LED)until the largest
                          > change occurs on the center sensor. However, every time it changes
                          position
                          > it has to dump the sensor history, hold still and acquire new "change"
                          data.
                          > In addition, for obvious reasons the sensors have to be "blind to change"
                          > while the sensors are being moved. A really smart head could do the matrix
                          > calculations to transfer the history of sensor data that remains in the
                          > field of view.
                          >
                          > Anyway you get the picture. For all but the simplest bright light tracking
                          > case, this software / memory map can not be easily adapted to the analog
                          > servo loop circuits of Beam heads.
                          >
                          > A different approach would use an array and a neural/nervous network to
                          > weight the outputs of the sensors and generate a 2 phase error signal to
                          > drive the motor servo loop. By AC coupling and rectifying the sensor
                          outputs
                          > (or equivalent) changes can be detected and "compared" to the changes in
                          > other sensors. Then a pulse is generated with a phase and duration
                          > proportional to angle and proximity of the sensor of the largest change to
                          > the sensor arry center. The effect of that will be to calculate the error
                          > signal and execute the correction after which the process repeats. During
                          > the execution phase that sensor "change" data is blocked. So the servo
                          loop
                          > process is no longer continuous but divided into discrete time steps.
                          >
                          > Next a few circuits that use a sensor array to turn beam head towards the
                          > brightests light in the field of view.
                          >
                          > wilf
                          >
                          > -----Original Message-----
                          > From: S. G.
                          > To: beam@egroups.com
                          > Sent: 11/29/00 8:28 PM
                          > Subject: [beam] Advanced Beam Head Technology Question
                          >
                          > Hi all,
                          > Greetings from Texas!
                          >
                          > I know a lot of you (including myself) have built powersmart beam heads,
                          > or
                          > maybe just simple bicore beam heads. I have enjoyed this technology,
                          > (thanks Wilf!) and have even sought (with Bruce's help) to incorporate
                          > one
                          > into a two motor walker. It's a work still in progress. ( ok ok, its on
                          > the
                          > shelf now, but I'll get there soon, I promise!)
                          >
                          > Now bear with me here, and please keep the flames regarding my stupidy
                          > on
                          > this matter to a minimum. :)
                          > Is it possible to do more than have the head turn towards a brighter
                          > source
                          > of light?
                          > The case I am thinking of is a head that turns towards any light change
                          > in
                          > the room. The theory of operation is that if something moves, then the
                          > light characteristics are going to change in that area. The bigger the
                          > change, the more apt the head is to look at it.
                          > Now, with a microprocessor the HEAD can determine when to move by can
                          > useing
                          > the LDRs as sort of a sensor which could generate a range of values
                          > based on
                          > the light level.
                          > This has been done with great results, as seen at:
                          > http://www.nwlink.com/~kevinro/lowresv.html
                          > which I have mentioned here before. BUT, I got to thinking, Why not
                          > use
                          > more LDRs in a vision configuration (such as in the URL mentioned above)
                          > and
                          > use lenses in the same way to bring light into focus, and the determine
                          > the
                          > change of light in the room in much the same way, but by using an analog
                          > (BEAM) head circuit?
                          > Seems to be the concept is close to the same, the encreased LDRs simply
                          > amplify the sensitivity, when they are brought into focus.
                          > The mechanics (configuration?) has already been done. Seems like a lot
                          > of
                          > work has already been accomplished for us. (again I'm referring to the
                          > above
                          > URL) What sort of modified power-smart head circuit would work on such
                          > a
                          > project?
                          > OR,
                          > am I simply way off base here? Seems to me that such a beast is
                          > possible.
                          > I am willing to invest atever $$ it takes to see if such a thing can
                          > happen,
                          > As I have some ideas for a bot like this.
                          > Comments?
                          > -Sparky
                          >
                          >
                          >
                          > To unsubscribe from this group, send an email to:
                          > beam-unsubscribe@egroups.com
                          >
                          >
                          >
                          >



                          To unsubscribe from this group, send an email to:
                          beam-unsubscribe@egroups.com
                        • Robert Morris
                          How about something like this, It s easier for me to explain with a diagram so... The output of the sensors (cds cells, etc.) feed into both a sample & hold
                          Message 12 of 14 , Nov 30, 2000
                          • 0 Attachment
                            How about something like this,
                            It's easier for me to explain with a diagram so...
                            The output of the sensors (cds cells, etc.) feed into both a sample & hold
                            mechanism and one side of a comparator. The output of the sample & hold
                            feeds into the other side of the comparator (Droidmaker's comparator?). The
                            comparator then feeds into a latch, which then feeds into the motor.
                            A 3 Nv ring with a very short TC turns on and off each section in sequence:
                            1. Turn on S&H, old sample is erased, new one stored.
                            2. Turn on Comparator, compare between sample and current sensor readings.
                            Store output in latch, 00=no movement, 01=movement left, 10=movement right.
                            3. Turn on motor, which moves according to stored latch information.
                            4. Repeat.

                            While the head may not move continuously, the timing might be fast enough
                            to appear nearly continuous.

                            Robert Morris

                            At 10:26 PM 11/29/00 -0800, you wrote:
                            >...
                            >To detect "movement" or "change", something else is needed: the array must
                            >be held steady with respect to the world or else, the way the array sees it,
                            >when the sensors are moving, the whole world would appear to move or change
                            >with respect to it. A histogram of variations in light levels on each sensor
                            >are accumulated and the servos are directed to adjust the center of view to
                            >the sensor with the largest changes (eg flashing LED)until the largest
                            >change occurs on the center sensor. However, every time it changes position
                            >it has to dump the sensor history, hold still and acquire new "change" data.
                            >In addition, for obvious reasons the sensors have to be "blind to change"
                            >while the sensors are being moved. A really smart head could do the matrix
                            >calculations to transfer the history of sensor data that remains in the
                            >field of view.
                            >
                            >Anyway you get the picture. For all but the simplest bright light tracking
                            >case, this software / memory map can not be easily adapted to the analog
                            >servo loop circuits of Beam heads.
                            >
                            >A different approach would use an array and a neural/nervous network to
                            >weight the outputs of the sensors and generate a 2 phase error signal to
                            >drive the motor servo loop. By AC coupling and rectifying the sensor outputs
                            >(or equivalent) changes can be detected and "compared" to the changes in
                            >other sensors. Then a pulse is generated with a phase and duration
                            >proportional to angle and proximity of the sensor of the largest change to
                            >the sensor arry center. The effect of that will be to calculate the error
                            >signal and execute the correction after which the process repeats. During
                            >the execution phase that sensor "change" data is blocked. So the servo loop
                            >process is no longer continuous but divided into discrete time steps.
                            >
                            >Next a few circuits that use a sensor array to turn beam head towards the
                            >brightests light in the field of view.
                            >
                            >wilf
                          • Robert Morris
                            Some further thoughts on my last post, The sensors could consist of 3 or more individual elements (LDR s, Cds, etc.) in odd numbers, i.e. 3,5,7,..., placed
                            Message 13 of 14 , Nov 30, 2000
                            • 0 Attachment
                              Some further thoughts on my last post,
                              The sensors could consist of 3 or more individual elements (LDR's, Cds,
                              etc.) in odd numbers, i.e. 3,5,7,..., placed along a single row. The Sample
                              & Hold circuit would store a reading from the sensors, and the comparator
                              would then compare the stored sample with a reading that it takes.
                              If there is no difference, then the head would not turn.
                              If the middle sensor changed, the head still would not turn because it is
                              already facing the movement.
                              If any of the left sensors changed, it would turn to the left.
                              If any of the right sensors changed, it would turn to the right.
                              It's goal would be to keep the changing sensor data coming from the middle.
                              The more sensors there are, the more accurate it would be (I think).

                              Robert Morris
                            • S. G.
                              . ... the hard part would be getting those sensors into focus. the LRVS seems to show how to do this. finnally a use for all those lenses after we take the
                              Message 14 of 14 , Nov 30, 2000
                              • 0 Attachment
                                .
                                > The more sensors there are, the more accurate it would
                                >be (I think).

                                the hard part would be getting those sensors into focus.
                                the "LRVS" seems to show how to do this.

                                finnally a use for all those lenses after we take the gearmotors off them!
                                -sparky
                              Your message has been successfully submitted and would be delivered to recipients shortly.