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

Mars Robot - Outline

Expand Messages
  • Martin Jay McKee
    All the talk about a mars compatible robot got me thinking (never a good thing). I m formulating a basic design in my head, and I m going to start putting it
    Message 1 of 17 , Jan 30, 2003
    • 0 Attachment
      All the talk about a mars compatible robot got me thinking (never a good
      thing). I'm formulating a basic design in my head, and I'm going to start
      putting it down on paper. I have a few things that I'm trying to
      incorporate into the design.

      1. Low cost
      2. Total Autonomy
      3. Damage compensation
      4. Basic intellegence (some idea of what it's supposed to do)

      First off... what else should I add to the list. At this point I'm totally
      disregarding the transportation to mars, and landing it there. I'm just
      thinking about the robot itself. What skills should it have.

      Second I've started building a basic list of materials. For insulation
      ROHACELL 31 foam, very low density (.032 g/cc) and inexpensive. For further
      insulation, aluminized mylar (space blankets). Between the two I think a
      fairly high amount of insulation should be possible (I'm going to do some
      heat transfer simulations at some point). Structure is proving to be less
      obvious, aluminum is on the list for it's weight, price, and workability.
      For propulsion I'm looking at muscle wires. Although their power
      consumption, and efficiency aren't ideal you can get them with transition
      temperatures as low as -200 degrees C, which means that they will have no
      problem with the temperatures. To protect the muscle wires, teflon tubing
      could easily be used, down to about -150 degrees C. As for control systems
      that's still very much up in the air. The electronics should be fully
      redundant, along with the power storage system. The ability to right itself
      would make the robot much more difficult to stop so it will need to be able
      to tell which way is up.

      Anyway that's what I got so far. Not much really, but it's a start.

      Martin Jay McKee
    • Ori
      A snickable design would be good. You know how socket pins come in those snickable strips? Well, a bot like that. Broken apart, it can still function. Mark
      Message 2 of 17 , Jan 30, 2003
      • 0 Attachment
        A "snickable" design would be good. You know how socket pins come in those
        "snickable" strips? Well, a bot like that. Broken apart, it can still
        function.

        Mark mentioned that if his electronics would have been properly arranged,
        Spyder could be broken into four parts and keep on running. That makes
        sense, because there are just four quadcores linked with resistors. if one
        of the quadrants breaks off, then the loop of the quadcores is disconnected,
        but the broken-off leg (assuming it has it's own power) would keep on going,
        and the rest of the bot would also keep on going.

        Why is that important? Well, NASA knows that if one of their Mars bots
        breaks, that is fatal. End of mission. The media (nor myself - hey, how many
        inches in a centimeter? :) would never let them forget it. If the bot could
        keep on going if the unexpected happens, then there is automatically a
        serious advantage.

        Ori

        ----- Original Message -----
        From: "Martin Jay McKee" <MartinJayMcKee@...>
        To: "BEAM" <beam@yahoogroups.com>
        Sent: Thursday, January 30, 2003 6:38 PM
        Subject: [beam] Mars Robot - Outline


        > All the talk about a mars compatible robot got me thinking (never a good
        > thing). I'm formulating a basic design in my head, and I'm going to start
        > putting it down on paper. I have a few things that I'm trying to
        > incorporate into the design.
        >
        > 1. Low cost
        > 2. Total Autonomy
        > 3. Damage compensation
        > 4. Basic intellegence (some idea of what it's supposed to do)
        >
        > First off... what else should I add to the list. At this point I'm
        totally
        > disregarding the transportation to mars, and landing it there. I'm just
        > thinking about the robot itself. What skills should it have.
        >
        > Second I've started building a basic list of materials. For insulation
        > ROHACELL 31 foam, very low density (.032 g/cc) and inexpensive. For
        further
        > insulation, aluminized mylar (space blankets). Between the two I think a
        > fairly high amount of insulation should be possible (I'm going to do some
        > heat transfer simulations at some point). Structure is proving to be less
        > obvious, aluminum is on the list for it's weight, price, and workability.
        > For propulsion I'm looking at muscle wires. Although their power
        > consumption, and efficiency aren't ideal you can get them with transition
        > temperatures as low as -200 degrees C, which means that they will have no
        > problem with the temperatures. To protect the muscle wires, teflon tubing
        > could easily be used, down to about -150 degrees C. As for control
        systems
        > that's still very much up in the air. The electronics should be fully
        > redundant, along with the power storage system. The ability to right
        itself
        > would make the robot much more difficult to stop so it will need to be
        able
        > to tell which way is up.
        >
        > Anyway that's what I got so far. Not much really, but it's a start.
        >
        > Martin Jay McKee
        >
        >
        > To unsubscribe from this group, send an email to:
        > beam-unsubscribe@egroups.com
        >
        >
        >
        > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
        >
        >
        >
      • Eric Seale <eric_seale@pobox.com>
        Ori, ... Keep in mind that there s a difference between something that keeps on going and something that s actually productive (I ll just write it off as
        Message 3 of 17 , Jan 30, 2003
        • 0 Attachment
          Ori,

          --- In beam@yahoogroups.com, "Ori" <sbarbut@s...> wrote:
          > A "snickable" design would be good. You know how socket pins come in those
          > "snickable" strips? Well, a bot like that. Broken apart, it can still
          > function.
          >
          > Mark mentioned that if his electronics would have been properly arranged,
          > Spyder could be broken into four parts and keep on running. That makes
          > sense, because there are just four quadcores linked with resistors. if one
          > of the quadrants breaks off, then the loop of the quadcores is disconnected,
          > but the broken-off leg (assuming it has it's own power) would keep on going,
          > and the rest of the bot would also keep on going.

          Keep in mind that there's a difference between something that "keeps on going" and something that's actually productive (I'll just write it off as another example of Mark's sometimes overly optimistic salesmanship). There are also more-likely failure scenarios for a rover than a broken leg (unless, of course, you make your legs "snickable" -- increasing the likelihood of *that particular* failure).

          > Why is that important? Well, NASA knows that if one of their Mars bots
          > breaks, that is fatal. End of mission. The media (nor myself - hey, how many
          > inches in a centimeter? :) would never let them forget it. If the bot could
          > keep on going if the unexpected happens, then there is automatically a
          > serious advantage.

          Yep, and accordingly all sorts of effort is expended on making sure they are robust to component failures (the upcoming MER rovers are a bit of an oddity in this -- each has "single string" components, so if anything breaks on one, the backup is the other rover). As for the '98 Orbiter, well it's tough to make anything that'll survive an accidental pass through the atmosphere at faster than orbital speed.

          It's also rarely discussed, but basically you can never really protect against the "unexpected" (your "snickable" walker is protected against failures that break off a leg -- thus, by definition, that failure is no longer "unexpected"). What you *can* do is quantify the things you're most likely to be up against, then protect your 'bot from them. There are some things you'll just never be able to protect against (being crushed by a giant asteroid, random black holes, hostile aliens) -- but you make darned sure you protect against anything even vaguely likely.

          Eric
        • tony stutsman
          i know this is going to sound stupid, okay fine this is going to sound really stupid but, i think we should forget about mars and start working towards the
          Message 4 of 17 , Jan 30, 2003
          • 0 Attachment

            i know this is going to sound stupid, okay fine this is going to sound really stupid but, i think we should forget about mars and start working towards the moon, besides its closer and there for won't take two years for the robot to land, if you can't tell i REALLY want to do something like this, so i think a group of us who like myself which REALLY wants to do something like this should start designing all that we would need to do it, like for one a rockit, which i would think is going to be the hard part, then design and build the radio or whatever that could send pics back for all to see and as for the robot design i was thinking  about a walker that had the legs on top and bottom so it would matter if it got turned over, but anywho, what do you guys think? can we send a beam robot to the moon?........damn i feel like a geek for saying that



            Do you Yahoo!?
            Yahoo! Mail Plus - Powerful. Affordable. Sign up now
          • Bruce Robinson
            ... I m afraid this is the sort of approach that leads to so many embarassing failures. By all means, take precautions against the known hazards. Just don t
            Message 5 of 17 , Jan 30, 2003
            • 0 Attachment
              "Eric Seale " wrote:
              >
              > ... It's also rarely discussed, but basically
              > you can never really protect against the
              > "unexpected" (your "snickable" walker is
              > protected against failures that break off a
              > leg -- thus, by definition, that failure is no
              > longer "unexpected"). What you *can* do is
              > quantify the things you're most likely to be
              > up against, then protect your 'bot from them.
              > There are some things you'll just never be
              > able to protect against (being crushed by a
              > giant asteroid, random black holes, hostile
              > aliens) -- but you make darned sure you protect
              > against anything even vaguely likely.

              I'm afraid this is the sort of approach that leads to so many
              embarassing failures. By all means, take precautions against the known
              hazards. Just don't build a robot that can only respond to an expected
              set of conditions. Rather, try to make your robot a set of feedback
              systems. Each system carries out a task by attaining an equilibrium
              state with its environment. When the system is disturbed (by the
              expected OR the unexpected), it responds by establishing a new
              equilibrium state that is very close to the target condition. It's the
              equilibrium condition that actually does the task.

              This is the sort of system that MWT was trying do create. He succeeded
              to a limited degree and I think there is a great deal of potential for
              much more complex systems. We just haven't got there yet. It's a lot
              easier to assemble some motors and levers and write a program that
              "makes" the assembly move. We even emulate the feedback system in
              software. But that leaves the system vulnerable to processor failure
              (soft or hard). It also seems to be darn easy to program a feedback loop
              that locks up when the system gets a little too far out of it's "normal"
              state.

              Of course it isn't realistic to expect this kind of system to survive an
              orbital accident -- that's a bit extreme. Or is it? Maybe it isn't after
              all. The planets manage to stay more or less in place, using exactly the
              sort of feedback system I've been proposing. Maybe the problem is we're
              trying to steer the darn thing into exactly the right position.

              Bruce
              (the dissident :)
            • Bruce Robinson
              ... In order of difficulty (starting with the hard part). 1) Landing on the moon, in one piece and still functional. 2) Getting to the moon from earth orbit.
              Message 6 of 17 , Jan 30, 2003
              • 0 Attachment
                tony stutsman wrote:
                >
                > ... can we send a beam robot to the moon?

                In order of difficulty (starting with the hard part).

                1) Landing on the moon, in one piece and still functional.
                2) Getting to the moon from earth orbit.
                3) Actually being able to move around on the moon.
                4) Building the darn thing.
                5) Getting into earth orbit.

                Bruce
              • Ori
                ... going, ... going and something that s actually productive (I ll just write it off as another example of Mark s sometimes overly optimistic salesmanship).
                Message 7 of 17 , Jan 31, 2003
                • 0 Attachment
                  > > but the broken-off leg (assuming it has it's own power) would keep on
                  going,
                  > > and the rest of the bot would also keep on going.
                  >
                  > Keep in mind that there's a difference between something that "keeps on
                  going" and something that's actually productive (I'll just write it off as
                  another example of Mark's sometimes overly optimistic salesmanship).

                  OK, salesman tactics do show up in those claims, but perhaps a bot *other*
                  than a normal BEAM walker could do it? Here are two ideas:

                  A snakebot, where each segment has it's own solar panel, gearmotor, and
                  quadcore (I'm going for a namechange here, no more monocore :). The
                  quadcores are all linked with the one in front of it with resistors. If the
                  segments break apart, then as long as more than 2 segments remain in each
                  section, then it keeps on going. And it is still 'actually productive'

                  A similar idea, but with legs. Make a millipede. Each segment has a
                  gearmotor with the output shaft pointing up, and back slightly. Each segment
                  also has a phototropic bicore (the side with more light doesn't step as
                  far). The bicores are also linked, from one segment to the next, with
                  resistors. Each motor shaft has a pair of legs attached, one on either side,
                  and it walks much like a two motor walker, except there are a lot of
                  segments. Each segment is connected to the adjacent segments with a
                  mechanism that can twist (to a limit), and allow up, down, left, and right
                  motion (also limited). This bot could also break apart, and if you have two
                  segments in a section, it's a fairly basic 2 motor walker. More segments
                  allow for more leg contacts, which means the bot would be better balanced.

                  So, two ideas of bots that can break, and still work - and be productive.

                  There are also more-likely failure scenarios for a rover than a broken leg
                  (unless, of course, you make your legs "snickable" -- increasing the
                  likelihood of *that particular* failure).

                  I didn't mean doing something like a preforated cut between each leg, to
                  make it break easily :) I think that by making a bot "snickable," some other
                  possible failures are taken care of. For one example, what happens if there
                  is an IC failure? If you would have put together something like Spyder with
                  two microcores on one IC, then you loose two legs. If you made the bot to
                  have seperate power (ground lines connected for reference), seperate
                  electronics, etc., then you only loose one leg.

                  > It's also rarely discussed, but basically you can never really protect
                  against the "unexpected" (your "snickable" walker is protected against
                  failures that break off a leg -- thus, by definition, that failure is no
                  longer "unexpected").

                  :) Good point, Eric.

                  >What you *can* do is quantify the things you're most likely to be up
                  against, then protect your 'bot from them. There are some things you'll
                  just never be able to protect against (being crushed by a giant asteroid,
                  random black holes, hostile aliens)

                  I think that the aliens might respond to a robot-mounted tazer...

                  Ori
                • Bill Bigge
                  A snakebot, where each segment has it s own solar panel, gear motor, and quadcore (I m going for a namechange here, no more monocore :). The quadcores are all
                  Message 8 of 17 , Jan 31, 2003
                  • 0 Attachment
                    Message
                    A snakebot, where each segment has it's own solar panel, gear motor, and
                    quadcore (I'm going for a namechange here, no more monocore :). The
                    quadcores are all linked with the one in front of it with resistors. If the
                    segments break apart, then as long as more than 2 segments remain in each
                    section, then it keeps on going. And it is still 'actually productive'
                     
                    What defines 'productive' here, is it just moving around or will the bot do anything else?  The only reason for sensong bots to mars at the moment (apart from for our own entertainment) is to do science so to be 'productive' the robot will have to perform some task, perhaps sampling the atmosphere as it travels along.
                     
                    If your task is to do science, or indeed anything else that we want to know about here on earth, you then need a way to get the data back which means radio transmitters or lasers.  Perhaps the 'simples' method, as far as the bot on mars goes, is to use an orbital relay station so the bot only has to transmit to orbit where the satellite will relay the signal to earth.
                     
                    So, two ideas of bots that can break, and still work - and be productive.
                    OK, if the bots task is to do some kind of science then a breakable bot would need to work in two ways:
                    A    -    It can loose a part of its body but still move around, carry its science instruments, and send a data stream to the orbital satellite.
                    B    -    It can break into several parts and each part can still move around - in this case in order to preserve its 'productiveness' each part needs to have the science tools and the data communication equipment - in this case it is almost worth defining the bot as a swarm and designing it to break up on command.  It could work like slime mould, moving around as a single entity and breaking up into a swarm when it finds something of interest.
                     
                    So.  Perhaps we need a 'task' for a robot, unless everyone is happy with entertainment.  Here are a couple of suggestions:
                     
                    A bot that can carry a payload - The bot is designed to get around and survive on mars, the payload could be a science package or whatever and it rides around on the robot (the horse and rider model)
                     
                    A bot with a camera - entertainment really, it just runs around on mars sending back a constant video stream of what it sees for us to look at and feel proud of.
                     
                    Bill
                    -----------------
                    Bill Bigge
                    Autonomous Systems Lab
                    Cognitive and Computing Sciences
                    Pevensey-II
                    Sussex University
                    Brighton
                    +44 (0)1273 877621
                    -----------------
                    www.cogs.susx.ac.uk
                     
                    -----Original Message-----
                    From: Ori [mailto:sbarbut@...]
                    Sent: 31 January 2003 15:49
                    To: beam@yahoogroups.com
                    Subject: Re: [beam] Re: Mars Robot - Outline

                    > > but the broken-off leg (assuming it has it's own power) would keep on
                    going,
                    > > and the rest of the bot would also keep on going.
                    >
                    > Keep in mind that there's a difference between something that "keeps on
                    going" and something that's actually productive (I'll just write it off as
                    another example of Mark's sometimes overly optimistic salesmanship).

                    OK, salesman tactics do show up in those claims, but perhaps a bot *other*
                    than a normal BEAM walker could do it? Here are two ideas:

                    A snakebot, where each segment has it's own solar panel, gearmotor, and
                    quadcore (I'm going for a namechange here, no more monocore :). The
                    quadcores are all linked with the one in front of it with resistors. If the
                    segments break apart, then as long as more than 2 segments remain in each
                    section, then it keeps on going. And it is still 'actually productive'

                    A similar idea, but with legs. Make a millipede. Each segment has a
                    gearmotor with the output shaft pointing up, and back slightly. Each segment
                    also has a phototropic bicore (the side with more light doesn't step as
                    far). The bicores are also linked, from one segment to the next, with
                    resistors. Each motor shaft has a pair of legs attached, one on either side,
                    and it walks much like a two motor walker, except there are a lot of
                    segments. Each segment is connected to the adjacent segments with a
                    mechanism that can twist (to a limit), and allow up, down, left, and right
                    motion (also limited). This bot could also break apart, and if you have two
                    segments in a section, it's a fairly basic 2 motor walker. More segments
                    allow for more leg contacts, which means the bot would be better balanced.

                    So, two ideas of bots that can break, and still work - and be productive.

                      There are also more-likely failure scenarios for a rover than a broken leg
                    (unless, of course, you make your legs "snickable" -- increasing the
                    likelihood of *that particular* failure).

                    I didn't mean doing something like a preforated cut between each leg, to
                    make it break easily :) I think that by making a bot "snickable," some other
                    possible failures are taken care of. For one example, what happens if there
                    is an IC failure? If you would have put together something like Spyder with
                    two microcores on one IC, then you loose two legs. If you made the bot to
                    have seperate power (ground lines connected for reference), seperate
                    electronics, etc., then you only loose one leg.

                    > It's also rarely discussed, but basically you can never really protect
                    against the "unexpected" (your "snickable" walker is protected against
                    failures that break off a leg -- thus, by definition, that failure is no
                    longer "unexpected").

                    :) Good point, Eric.

                    >What you *can* do is quantify the things you're most likely to be up
                    against, then protect your 'bot from them.  There are some things you'll
                    just never be able to protect against (being crushed by a giant asteroid,
                    random black holes, hostile aliens)

                    I think that the aliens might respond to a robot-mounted tazer...

                    Ori


                    To unsubscribe from this group, send an email to:
                    beam-unsubscribe@egroups.com



                    Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
                  • Richard Caudle <frankendaddy@yahoo.com>
                    All it takes to accomplish crazy things is a group of motivated crazy people. I think that it s at least an interesting exercise. There are the right mix of
                    Message 9 of 17 , Jan 31, 2003
                    • 0 Attachment
                      All it takes to accomplish crazy things is a group of motivated crazy
                      people. I think that it's at least an interesting exercise. There
                      are the right mix of talents here on the list that have the
                      possibility of coming up with a viable robotic space exploration
                      system. Given enough time and work, we might be able to get it on
                      the space shuttle at least.

                      Sounds better to me than urine-based paints!

                      Richard

                      p.s. Geeks Rule! You either are one or work for one.
                    • Sathe Dilip
                      A desert! They try to find a close match for the environment and test the rovers like this: http://www.flatoday.com/space/explore/stories/1999/050299b.htm
                      Message 10 of 17 , Jan 31, 2003
                      • 0 Attachment
                        A desert! They try to find a close match for the environment and test
                        the rovers like this:

                        http://www.flatoday.com/space/explore/stories/1999/050299b.htm

                        There is some footage of a actual landing here - with the rover roaming
                        around afterwards (probably computer generated):

                        http://www.space.com/media/s010717_2002_rovers_2.mov

                        Dilip
                        -----------------------------------

                        Wilf Rigter wrote:

                        > I think Mars and the Moon are great objectives and no doubt some of
                        > you will be part of the team that makes it happen. Now you know it
                        > won't be easy; just designing a rover that will survive in your own
                        > home is a non-trivial project. But yes, aim high to design and build a
                        > rover for Mars. Of course first you must prove that your rover will
                        > survive and carry out missions in the harshest unstructured
                        > environments right here on earth (on top of a mountain maybe?) .
                      • J Wolfgang Goerlich <jwgoerlich@hotmail.
                        ... I have been following this and the fountain of youth posts. Also, I have been reading thru the past year s posts in order to get a feel for this group.
                        Message 11 of 17 , Jan 31, 2003
                        • 0 Attachment
                          > All the talk about a mars compatible robot got me thinking ...

                          I have been following this and the fountain of youth posts. Also, I
                          have been reading thru the past year's posts in order to get a feel
                          for this group. With that as my background, I am going to stick my
                          neck out, and maybe ruffle a few feathers.

                          I have read about Mars rovers, and about unbreakable Beam robots. I
                          have also read about ecosystems, robotic reproduction, and the like.
                          These are all very exciting topics, and always generate a lot of
                          interest. Interest is a good thing, driving thought and innovation.

                          My concern about interest in general, and these topics in specific,
                          is that they are quite out of most of our reach. I, for one, am
                          limited to about $50 USD per week for my robotics hobby. That is
                          enough to play with, but certainly not enough to get me to the Moon,
                          or Mars, or to build a robotic factory. Thus, for me at least, these
                          topics are way out of reach.

                          Out of reach is a problem, because it means that we will not be able
                          to do it. Lots of thinking, and discussions, will then ultimatetly
                          result in no new robotics. Therefore, my concern really is one of
                          practicality. It would seem that a lot of time is being spent on
                          things that we cannot build, while little time is being spent on
                          things that we can (albeit things that are less glamorous).

                          And what are these things? Bruce Robinson's learning nets are one.
                          Marty Vulk's outdoors turbots are another. Heck, just about any
                          innovative circuit from Wilf would be interesting. And this gets to
                          my questions: How many Beam robots have been built with a basic
                          ability to learn (a la Bruce Robinson's design, or others)? How many
                          robots have been built that can survive outdoors, on Earth?

                          In short, I would propose we start out with simplier goals that we
                          can accomplish. Then, thru accomplishing these goals, build upon them
                          to create ever complex designs. Tilden argued that we are the agent
                          for robotic evolution, right? Well, taking the B from Beam, in
                          biology, many, many simplier forms of life had to evolve before
                          complex mammals. Likewise, many, many simplier forms of robots will
                          have to be evolved before we can shoot for the Moon, Mars, outer
                          space, et cetera.

                          Well, there is my two cents worth. Thank you,
                          J Wolfgang Goerlich


                          Related Links:

                          Bruce Robinson's Learning Robots article
                          http://www3.telus.net/rfws/beam/learn/lr_00.html

                          My Turbots in the backyard post:
                          http://groups.yahoo.com/group/beam/message/33154

                          Luther Burrell's Backyard RJP post:
                          http://groups.yahoo.com/group/beam/message/22449
                        • Eric Seale <eric_seale@pobox.com>
                          ... You ll want to think about science / sensing. After all, sending the bot off to Mars won t do much good if it just wanders around without taking pictures
                          Message 12 of 17 , Jan 31, 2003
                          • 0 Attachment
                            --- In beam@yahoogroups.com, "Martin Jay McKee" <MartinJayMcKee@a...> wrote:
                            > All the talk about a mars compatible robot got me thinking...
                            > ...I have a few things that I'm trying to
                            > incorporate into the design.
                            >
                            > 1. Low cost
                            > 2. Total Autonomy
                            > 3. Damage compensation
                            > 4. Basic intellegence (some idea of what it's supposed to do)
                            >
                            > First off... what else should I add to the list. At this point I'm totally
                            > disregarding the transportation to mars, and landing it there. I'm just
                            > thinking about the robot itself. What skills should it have.

                            You'll want to think about science / sensing. After all, sending the 'bot off to Mars won't do much good if it just wanders around without taking pictures or collecting some kind of interesting data. Maybe something simple -- like foot-pad mounted sensors that react to one given mineral. You could send multiple 'bots, each sensitive to a different material; as they march across the surface, you can tell what's located where.

                            You'll also want to think about telemetry -- how do you get news of your robot's interesting discoveries back home? This is usually the toughest hurdle for small robots -- it costs a given amount of power to transmit (via RF) regardless of whether your 'bot is big or small. Small 'bots just plain have a harder time coming up with the power to support telemetry. Some folks are now researching passive (one-way) telecom -- say put a corner-cube reflector on your robot that you can tune / detune to send telemetry. This requires a laser, telescope, and some pretty fancy signal processing logic on an orbiter to act as your relay. If you're using an orbital relay, you also have to deal with the fact that you can only transmit for a short period of time in a session, and your available sessions will be limited by your relay's orbit.

                            >...As for control systems
                            > that's still very much up in the air. The electronics should be fully
                            > redundant, along with the power storage system.

                            Redundancy is generally good -- the thing you'll have to be careful about is that it's easy to make a system *less reliable* when you add the redundancy (since the circuitry to handle the redundancy adds complexity, which adds failure modes).

                            >...The ability to right itself...

                            This is good for a small 'bot.

                            > would make the robot much more difficult to stop so it will need to be able
                            > to tell which way is up.

                            Lots of "tilt" switches are available for this.

                            Eric
                          • Bruce Robinson
                            ... At this stage in Mars exploration, an experiment in mobility is itself a productive task. The rovers are all wheeled robots, with the wheels mounted on
                            Message 13 of 17 , Jan 31, 2003
                            • 0 Attachment
                              Bill Bigge wrote:

                              > What defines 'productive' here, is it
                              > just moving around or will the bot do
                              > anything else? The only reason for
                              > sensong bots to mars at the moment
                              > (apart from for our own entertainment)
                              > is to do science so to be 'productive'
                              > the robot will have to perform some task,
                              > perhaps sampling the atmosphere as it
                              > travels along.

                              At this stage in Mars exploration, an experiment in mobility is itself a
                              productive task. The rovers are all wheeled robots, with the wheels
                              mounted on movable arms. Their purpose is to move a "scientific package"
                              around the surface of the planet, to learn more about the planet. The
                              wheeled design is used because it seems to be the "best" design (i.e.,
                              least likely to fail) for carrying the package (based on experiments on
                              Earth).

                              Logically, you'd expect to first try several experiments on Mars to see
                              what type of motive unit worked best on the actual planet. Then you'd
                              try to incorporate a scientific package into the better designs, and
                              only then would you send the "real" experimental device.

                              Testing motive units on Mars first makes a lot of sense. Have a base
                              station that can communicate with earth and then deploy dozens of
                              (small) different types of robots that only need to transmit information
                              tothe base station. You could have Tilden walkers, snakebots, wheeled
                              robots, crawlers, hoppers, you name it. The expedition becomes one huge
                              experiment in "What's the best way to move around Mars?"

                              But try to get funding for something like that. Not a chance. Imagine a
                              senator trying to explain to the voters why he supported spending
                              hundreds of millions of dollars to send a bunch of "toys" to Mars to see
                              which one could get around better. No. We have to be "practical". That
                              leaves us with the remote possibility of having the main mission -- the
                              rover -- drop off a small package containing a self-extracting mobile
                              platform -- say a snakebot. Drop it off, have it self-activate after the
                              rover is out of the area, and just let it try to travel. Have the rover
                              come back every once in a while to see how it's doing. Minimal chance of
                              complicating the mission, no telemetry, no sensors beyond those it needs
                              for its own motion.

                              Even there, of course, you will have trouble getting support. Can't take
                              the slightest chance of messing up the "real" mission. Besides, what are
                              you setting out to do? You might end up proving the original mission
                              designers picked the wrong type of motion.

                              My little story about James Lovelock has an interesting connection to
                              Mars, and to funding for space exploration. When he applied for his
                              grant to build an instrument to measure halocarbons in the atmosphere,
                              he already had some impressive credentials. He had been hired by NASA in
                              1961 to work one instrumentation experiments on the first unmanned
                              flight to the moon. Not long afterward he was transferred to the section
                              that was designing instruments for the Viking lander, to detect life on
                              Mars. He soon realized that most of the scientists were basing their
                              "life detection" devices on their own very narrow view of what
                              constituted "life". He even showed that most of these devices would fail
                              to detect life on Earth if they landed, say, in the antartic. Then he
                              went on to show a much more reliable way to detect life on a planet.
                              Using existing instrumentation on Earth, he was able to demonstrate that
                              life did not exist on Mars, and could not have existed their for
                              millions of years, if at all.

                              Right. Just what his bosses wanted to hear. The funding for Viking had
                              been obtained partly for the purpose of determining whether life existed
                              on Mars, and here's this guy who just proved -- more than a decade
                              before the actual flight was scheduled -- that it didn't. So this little
                              scientific tidbit was suppressed so as not to jeopordize the first
                              landing on Mars. (Not that Lovelock objected).

                              So there's another twist on getting a BEAM robot into space, or onto the
                              surface of a planet. You have to have a purpose that will (A) "grab" the
                              funders, (B) interest the scientists, and (C) not threaten the success
                              of the main mission.

                              Bruce
                            • Martin Jay McKee
                              I agree, for myself I like the idea of designing a robot for Mars (or the moon) purly because it s a major challenge. Most of the parts of the shuttle, ISS,
                              Message 14 of 17 , Jan 31, 2003
                              • 0 Attachment
                                I agree, for myself I like the idea of designing a robot for Mars (or the moon) purly because it's a major challenge.  Most of the parts of the shuttle, ISS, etc. that were tested on Earth are completely different than the version that was sent up simply because of the change in environment (temperature, radiation, gravity).  Actually building the robot, exactly like it would be, isn't even on my list, I'm an engineer, I like designing things.
                                 
                                Besides, any new ideas that crop up in the process of designing a robot for a harsh environment, like Mars, can be adapted for Earth as well.  Pushing technology, and our knowledge up to and past where it is now is the way that breakthroughs are made.
                                 
                                Martin Jay McKee
                                -----Original Message-----
                                From: Wilf Rigter [mailto:wrigter@...]
                                Sent: Monday, March 03, 2003 00:51
                                To: beam@yahoogroups.com
                                Subject: Re: [beam] Re: Mars Robot - Outline

                                I think Mars and the Moon are great objectives and no doubt some of you will be part of the team that makes it happen. Now you know it won't be easy;  just designing a rover that will survive in your own home is a non-trivial project. But yes, aim high to design and build a rover for Mars. Of course first you must prove that your rover will survive and carry out missions in the harshest unstructured environments right here on earth (on top of a mountain maybe?) .  Next test it under the same severe temperature and low pressure conditions as Mars in an environmental chamber.  If it can survive all that, you will already have accomplished a remarkable feat, indeed you will be famous and  ready to take the next step or should I say, the extra few million miles, to test the full  ET version of the rover on the red planet itself.
                                 
                                wilf
                                ----- Original Message -----
                                Sent: Thursday, January 30, 2003 9:31 PM
                                Subject: Re: [beam] Re: Mars Robot - Outline

                                i know this is going to sound stupid, okay fine this is going to sound really stupid but, i think we should forget about mars and start working towards the moon, besides its closer and there for won't take two years for the robot to land, if you can't tell i REALLY want to do something like this, so i think a group of us who like myself which REALLY wants to do something like this should start designing all that we would need to do it, like for one a rockit, which i would think is going to be the hard part, then design and build the radio or whatever that could send pics back for all to see and as for the robot design i was thinking  about a walker that had the legs on top and bottom so it would matter if it got turned over, but anywho, what do you guys think? can we send a beam robot to the moon?........damn i feel like a geek for saying that



                                To unsubscribe from this group, send an email to:
                                beam-unsubscribe@egroups.com



                                Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
                              • Eric Seale <eric_seale@pobox.com>
                                ... known ... expected ... OK, I ll bite. Before you ll get approval to fly a robot to Mars or the Moon or wherever, you ll need to validate its performance
                                Message 15 of 17 , Jan 31, 2003
                                • 0 Attachment
                                  --- In beam@yahoogroups.com, Bruce Robinson <Bruce_Robinson@t...>
                                  wrote:
                                  > "Eric Seale " wrote:
                                  > > ... It's also rarely discussed, but basically
                                  > > you can never really protect against the
                                  > > "unexpected" (your "snickable" walker is
                                  > > protected against failures that break off a
                                  > > leg -- thus, by definition, that failure is no
                                  > > longer "unexpected"). What you *can* do is
                                  > > quantify the things you're most likely to be
                                  > > up against, then protect your 'bot from them.
                                  > > There are some things you'll just never be
                                  > > able to protect against (being crushed by a
                                  > > giant asteroid, random black holes, hostile
                                  > > aliens) -- but you make darned sure you protect
                                  > > against anything even vaguely likely.
                                  >
                                  > I'm afraid this is the sort of approach that leads to so many
                                  > embarassing failures. By all means, take precautions against the
                                  known
                                  > hazards. Just don't build a robot that can only respond to an
                                  expected
                                  > set of conditions.

                                  OK, I'll bite. Before you'll get approval to fly a robot to Mars or
                                  the Moon or wherever, you'll need to validate its performance in a
                                  variety of ways (safety, reliability, etc.). If not the folks paying
                                  for your robot (via grant or loan or whatever), at least the folks
                                  giving you your flight will expect some assurances. How do you prove
                                  it'll do something beneficial in "unexpected" conditions? How do you
                                  prove it won't do something hazardous or suicidal in "unexpected"
                                  conditions? How can you possibly define a test for "unexpected"
                                  conditions (shades of Schroedinger's cat)?

                                  If you can't / don't quantify the conditions your robot is capable of
                                  handling, it'll be tough for you to find anybody that wants to mess
                                  with it (aerospace, thanks to the high visibility of any catastrophic
                                  failures, is a conservative business).

                                  >... Rather, try to make your robot a set of feedback
                                  > systems. Each system carries out a task by attaining an equilibrium
                                  > state with its environment. When the system is disturbed (by the
                                  > expected OR the unexpected), it responds by establishing a new
                                  > equilibrium state that is very close to the target condition. It's
                                  the
                                  > equilibrium condition that actually does the task.

                                  Absolutely -- although (maybe it's just a disagreement over
                                  semantics) I'd argue that this is still a bounded problem in a finite
                                  state space, and still leaves little room for the unexpected.

                                  > ...It's a lot
                                  > easier to assemble some motors and levers and write a program that
                                  > "makes" the assembly move. We even emulate the feedback system in
                                  > software. But that leaves the system vulnerable to processor failure
                                  > (soft or hard).

                                  True, decentralized systems can be much more robust than centralized
                                  ones (and processors make it awfully easy to centralized). Do you
                                  think that emergent behavior really has to come only from hardware,
                                  though?

                                  >...It also seems to be darn easy to program a feedback loop
                                  > that locks up when the system gets a little too far out of it's
                                  "normal"
                                  > state.

                                  True, but it's also easy to make non-robust hardware (it's easy to do
                                  anything badly).

                                  > Of course it isn't realistic to expect this kind of system to
                                  survive an
                                  > orbital accident -- that's a bit extreme. Or is it? Maybe it isn't
                                  after
                                  > all. The planets manage to stay more or less in place, using
                                  exactly the
                                  > sort of feedback system I've been proposing. Maybe the problem is
                                  we're
                                  > trying to steer the darn thing into exactly the right position.

                                  Eh.... I'd disagree that orbital dynamics involves anything
                                  resembling BEAM-style feedback. More to the point, though, some
                                  problems are just not too forgiving of error -- orbital insertion,
                                  brain surgery...

                                  Eric
                                • Bruce Robinson
                                  ... You don t. You prove it will do something beneficial in expected conditions. That s what the sponsors want. The fact that it will handle other conditions
                                  Message 16 of 17 , Jan 31, 2003
                                  • 0 Attachment
                                    "Eric Seale " wrote:
                                    >
                                    > ... How do you prove it'll do something
                                    > beneficial in "unexpected" conditions? ...

                                    You don't. You prove it will do something beneficial in expected
                                    conditions. That's what the sponsors want. The fact that it will handle
                                    other conditions as well is a bonus.

                                    > ... How do you prove it won't do something
                                    > hazardous or suicidal in "unexpected"
                                    > conditions? ...

                                    You don't need to. If they haven't thought of the condition, they can't
                                    very well complain if your robot can't handle it.

                                    > ... How can you possibly define a test for
                                    > "unexpected" conditions (shades of Schroedinger's
                                    > cat)?

                                    You don't need to.

                                    The point being, you devise a robot able to handle the unexpected. Then
                                    you throw some random conditions at it that you don't expect it to
                                    handle and see how robust the system is. When you figure you have a
                                    pretty robust system that can handle all the "expected" conditions, you
                                    send it up. A few years later when the darn thing is still going, into
                                    places the sponsors thought were impossible to negotiate, then people
                                    will start to say, "Hey, those guys design pretty good robots."

                                    Of course, this can backfire. Such as when someone sends up a really,
                                    really sophisticated robot 2 years later and your robot ends up
                                    wandering over to bail it out when it breaks down :)

                                    > If you can't / don't quantify the conditions
                                    > your robot is capable of handling, it'll be
                                    > tough for you to find anybody that wants to mess
                                    > with it ...

                                    But you do quantify the conditions -- the ones you know about and of
                                    course the ones they mandate for their mission.

                                    >> ... Rather, try to make your robot a set of
                                    >> feedback systems. Each system carries out a
                                    >> task by attaining an equilibrium state with
                                    >> its environment. When the system is disturbed
                                    >> (by the expected OR the unexpected), it responds
                                    >> by establishing a new equilibrium state that
                                    >> is very close to the target condition. It's
                                    >> the equilibrium condition that actually does
                                    >> the task.
                                    >
                                    > Absolutely -- although (maybe it's just a disagreement
                                    > over semantics) I'd argue that this is still a bounded
                                    > problem in a finite state space, and still leaves little
                                    > room for the unexpected.

                                    A single feedback subsystem may not leave much room for the unexpected.
                                    A system made up of many subsystems (and sub-subsystems) will certainly
                                    surprise you.

                                    > Do you think that emergent behavior really
                                    > has to come only from hardware, though?

                                    No. Emergent behaviour comes from systems.

                                    >> ... The planets manage to stay more or less
                                    >> in place, using exactly the sort of feedback
                                    >> system I've been proposing. ...
                                    >
                                    > Eh.... I'd disagree that orbital dynamics involves
                                    > anything resembling BEAM-style feedback. More to
                                    > the point, though, some problems are just not too
                                    > forgiving of error ...

                                    What's BEAM-style feedback? I'm talking about dynamic systems (which by
                                    definition incorporate feedback).

                                    When a problem is not too forgiving of error, you're often looking at
                                    the wrong problem.

                                    :)

                                    Bruce
                                  • Wilf Rigter
                                    I think Mars and the Moon are great objectives and no doubt some of you will be part of the team that makes it happen. Now you know it won t be easy; just
                                    Message 17 of 17 , Mar 2 10:51 PM
                                    • 0 Attachment
                                      I think Mars and the Moon are great objectives and no doubt some of you will be part of the team that makes it happen. Now you know it won't be easy;  just designing a rover that will survive in your own home is a non-trivial project. But yes, aim high to design and build a rover for Mars. Of course first you must prove that your rover will survive and carry out missions in the harshest unstructured environments right here on earth (on top of a mountain maybe?) .  Next test it under the same severe temperature and low pressure conditions as Mars in an environmental chamber.  If it can survive all that, you will already have accomplished a remarkable feat, indeed you will be famous and  ready to take the next step or should I say, the extra few million miles, to test the full  ET version of the rover on the red planet itself.
                                       
                                      wilf
                                      ----- Original Message -----
                                      Sent: Thursday, January 30, 2003 9:31 PM
                                      Subject: Re: [beam] Re: Mars Robot - Outline

                                      i know this is going to sound stupid, okay fine this is going to sound really stupid but, i think we should forget about mars and start working towards the moon, besides its closer and there for won't take two years for the robot to land, if you can't tell i REALLY want to do something like this, so i think a group of us who like myself which REALLY wants to do something like this should start designing all that we would need to do it, like for one a rockit, which i would think is going to be the hard part, then design and build the radio or whatever that could send pics back for all to see and as for the robot design i was thinking  about a walker that had the legs on top and bottom so it would matter if it got turned over, but anywho, what do you guys think? can we send a beam robot to the moon?........damn i feel like a geek for saying that


                                    Your message has been successfully submitted and would be delivered to recipients shortly.