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

memory

Expand Messages
  • lightninjim1
    First up i d like to say thanks to wilf for the welcome. Has anyone managed to come up with any type of long term memory system for beambots (by long term i
    Message 1 of 4 , Jan 29, 2007
    • 0 Attachment
      First up i'd like to say thanks to wilf for the welcome.

      Has anyone managed to come up with any type of long term memory system
      for beambots (by long term i mean more than a few cycles). I started
      thinking on this a couple of weeks ago, but as yet nothing. I've yet to
      build my first beambot but i've never been able to keep my head in
      the 'here and now', im always looking at the next step.
      Thing is i've thought of building a hybrid beam/PiC bot, but i hate
      programming and believe that given enough thought a solution will
      always present itself.
      Anyway enough of my rambling. Any feedback (Positive or Negative) will
      be greatly appreciated.
      Jim
    • mycroft2152
      Hi Jim, Take a look at the PICAXE microprocessors. They are very cheap and simple to program in BASIC. No programmer is needed, just a simple cable. They have
      Message 2 of 4 , Jan 29, 2007
      • 0 Attachment
        Hi Jim,

        Take a look at the PICAXE microprocessors. They are very cheap and
        simple to program in BASIC. No programmer is needed, just a simple
        cable. They have been discussed here before and have been used in
        SMARTBEAM circuits.

        They are a lot of fun.

        Myc

        --- In beam@yahoogroups.com, "lightninjim1" <lightninjim1@...> wrote:
        >
        > First up i'd like to say thanks to wilf for the welcome.
        >
        > Has anyone managed to come up with any type of long term memory
        system
        > for beambots (by long term i mean more than a few cycles). I started
        > thinking on this a couple of weeks ago, but as yet nothing. I've yet
        to
        > build my first beambot but i've never been able to keep my head in
        > the 'here and now', im always looking at the next step.
        > Thing is i've thought of building a hybrid beam/PiC bot, but i hate
        > programming and believe that given enough thought a solution will
        > always present itself.
        > Anyway enough of my rambling. Any feedback (Positive or Negative)
        will
        > be greatly appreciated.
        > Jim
        >
      • wilf_nv
        There are several options for long term memory. Some of them involve adding digital logic using discrete logic and/or RAM. Let me describe a pure beam long
        Message 3 of 4 , Jan 31, 2007
        • 0 Attachment
          There are several options for long term memory. Some of them involve
          adding digital logic using discrete logic and/or RAM.

          Let me describe a "pure" beam long term memory that uses beam
          flipflops made from simple cross-coupled inverters. For example, one
          group of four inverters of an HC240 makes two RS flip-flops with
          complementary outputs while the remaining four inverters and
          corresponding enable pin can buffer and tri-state the flip-flop
          outputs connected to a "beam control bus". These HC240 memory cells
          can use a coin battery to make them non-volatile long term memories,
          maintaining their memorized state during power down.

          Say 16 memory cells can store a combination of high or low states
          that, when enabled on the beam bus, are connected to RC networks that
          influence the timing of beam neurons. The gating of the memory cells
          can depend on a programmed sequence (counter) or on the state of
          external sensors. The programming of the memory can be as a result of
          trial and error or operator assisted.

          In general the basic behavior of the beam bot "horse" is hardwired
          but the "rider" memory can change the timing or polarity of
          the "horse" beam signals and indirectly influence the beam
          behavior.

          The NV memory project was an example of this using a Non Volatile
          SRAM and digital logic to "record" beam signals as a beam robots
          reacts to its environment. The designer can decide if a particular
          sequence was novel (for example backing out of a tight spot) and save
          it. Then later, these save beam signals can be played back, when
          addressed by the right a combination of external signal to make the
          beam bot repeat the same behavior.

          http://tech.groups.yahoo.com/group/beam/files/wilf_nv/Nv
          memorizer.gif

          A more versatile hybrid uses a beam "horse" and
          microcontroller "rider".
          Solarbotics several different uC chips in some of their designs. The
          Basic stamp is another possible solution but relatively expensive.
          For a couple of dollars, you can get a nice little basic
          microcontroller on a chip with an excellent free Basic editor and
          simulator, programmed with a simple serial cable for quick results,
          and the winner is … the Picaxe microcontroller from Rev Ed.

          wilf



          --- In beam@yahoogroups.com, "lightninjim1" <lightninjim1@...> wrote:
          >
          > First up i'd like to say thanks to wilf for the welcome.
          >
          > Has anyone managed to come up with any type of long term memory
          system
          > for beambots (by long term i mean more than a few cycles). I
          started
          > thinking on this a couple of weeks ago, but as yet nothing. I've
          yet to
          > build my first beambot but i've never been able to keep my head in
          > the 'here and now', im always looking at the next step.
          > Thing is i've thought of building a hybrid beam/PiC bot, but i hate
          > programming and believe that given enough thought a solution will
          > always present itself.
          > Anyway enough of my rambling. Any feedback (Positive or Negative)
          will
          > be greatly appreciated.
          > Jim
          >
        • lightninjim1
          ... that ... of ... save ... The ... I couldnt get the link to work even by pasting but i get what you mean. I ve had a look at the picaxe system and as its
          Message 4 of 4 , Feb 1 2:34 AM
          • 0 Attachment
            --- In beam@yahoogroups.com, "wilf_nv" <wrigter@...> wrote:
            >
            > There are several options for long term memory. Some of them involve
            > adding digital logic using discrete logic and/or RAM.
            >
            > Let me describe a "pure" beam long term memory that uses beam
            > flipflops made from simple cross-coupled inverters. For example, one
            > group of four inverters of an HC240 makes two RS flip-flops with
            > complementary outputs while the remaining four inverters and
            > corresponding enable pin can buffer and tri-state the flip-flop
            > outputs connected to a "beam control bus". These HC240 memory cells
            > can use a coin battery to make them non-volatile long term memories,
            > maintaining their memorized state during power down.
            >
            > Say 16 memory cells can store a combination of high or low states
            > that, when enabled on the beam bus, are connected to RC networks
            that
            > influence the timing of beam neurons. The gating of the memory cells
            > can depend on a programmed sequence (counter) or on the state of
            > external sensors. The programming of the memory can be as a result
            of
            > trial and error or operator assisted.
            >
            > In general the basic behavior of the beam bot "horse" is hardwired
            > but the "rider" memory can change the timing or polarity of
            > the "horse" beam signals and indirectly influence the beam
            > behavior.
            >
            > The NV memory project was an example of this using a Non Volatile
            > SRAM and digital logic to "record" beam signals as a beam robots
            > reacts to its environment. The designer can decide if a particular
            > sequence was novel (for example backing out of a tight spot) and
            save
            > it. Then later, these save beam signals can be played back, when
            > addressed by the right a combination of external signal to make the
            > beam bot repeat the same behavior.
            >
            > http://tech.groups.yahoo.com/group/beam/files/wilf_nv/Nv
            > memorizer.gif
            >
            > A more versatile hybrid uses a beam "horse" and
            > microcontroller "rider".
            > Solarbotics several different uC chips in some of their designs.
            The
            > Basic stamp is another possible solution but relatively expensive.
            > For a couple of dollars, you can get a nice little basic
            > microcontroller on a chip with an excellent free Basic editor and
            > simulator, programmed with a simple serial cable for quick results,
            > and the winner is … the Picaxe microcontroller from Rev Ed.
            >
            > wilf
            >
            >
            > Thanks wilf.
            I couldnt get the link to work even by pasting but i get what you
            mean.
            I've had a look at the picaxe system and as its programmed in basic
            it looks to be promising. I also took the opportunity to read some of
            your posts on the picaxe and eagerly await to hear how your
            programming (BEAM style) comes along.
            When i eventually get my two motor walker built and working ill be
            running a few experiments on that before progressing to the next
            stage. Still having problems finding chips to salvage, but i only
            just finished building my p.s.u. a couple of days ago so no major
            hurry at the moment. My wife thinks im spending far too long
            researching etc, and wants a share of my time too. Cant blame her
            really.
            Again thanks very much for the input, its greatly appreciated.
            Jim
          Your message has been successfully submitted and would be delivered to recipients shortly.