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

New Tricks

Expand Messages
  • winglabs_inc
    So I took out my old Hexbug Original out earlier today. I haven t messed with it in awhile, since I couldn t figure out quite how to fix it. Turns out, all I
    Message 1 of 7 , Aug 9, 2014
    • 0 Attachment

      So I took out my old Hexbug Original out earlier today. I haven't messed with it in awhile, since I couldn't figure out quite how to fix it. Turns out, all I had to do was bend the rear left leg out a little, and it was moving fine again.

       

      But I did notice something very interesting while letting it run on the foot of my bedframe. If it encountered the edge of the wood at approximately a 5 degree angle or less, it was able to turn away enough to avoid falling. I was quite surprised by this, and thusly I think I've uncovered yet another trick behind the hexapod gait.

       

      You would think that when one leg goes off the edge, that side of the bug produces less thrust, and so the bug turns toward the edge and falls to its [possible] doom. But instead the rule my Hexbug seems to obey is exactly the opposite. Given a little thought, I realized why this is so:

      When one leg loses contact with the surface, the robot's balance becomes distributed amongst only 5 legs. In order to maintain balance, 3 legs must always have traction. With only 5 legs, the middle leg on the affected side ends up being on the ground for longer, because without adequate support, the chassis sinks down as the middle leg tries to raise up. It finally leaves the ground and the rear leg takes over, because the weight then shifts forward to the foreleg on the other side, which is supposed to be raised. So now there are two legs down on the other side when only one should be, and the middle leg on the affected side being on the ground longer than it should be. So the ultimate result is: the side with one less leg produces MORE thrust, and the other side produces LESS thrust, so the chassis actually turns AWAY from the edge!

       

      This is pretty fascinating to me, because with some development, a chassis could be designed that takes advantage of this quirk to passively implement edge avoidance, without any added complexity to the circuit.

      When I get a chance, I could build a 5 motor hexapod walker to test this further. The optimal implementation of this principle would probably be to make the forelegs the longest, with the mid legs being the shortest.

       

      Enjoy, Connor.

    • Martin McKee
      Yes, a very interesting observation -- and great analysis! I have often been rather frustrated by my lack of deep mechanical competence ( some would argue
      Message 2 of 7 , Aug 10, 2014
      • 0 Attachment
        Yes, a very interesting observation -- and great analysis!  I have often been rather frustrated by my lack of deep mechanical competence ( some would argue that point because I can build almost anything, but understanding mechanics to the point of analysis and building it are two very different things ) as it seems that animal locomotion is at least as much about the proper mechanical design as it is about the control systems behind it. There is no doubt that careful design of a walker could lead both to eased control requirements and reduced power consumption.

        Martin Jay McKee


        On Sat, Aug 9, 2014 at 8:39 PM, connor_ramsey@... [beam] <beam@yahoogroups.com> wrote:
         

        So I took out my old Hexbug Original out earlier today. I haven't messed with it in awhile, since I couldn't figure out quite how to fix it. Turns out, all I had to do was bend the rear left leg out a little, and it was moving fine again.

         

        But I did notice something very interesting while letting it run on the foot of my bedframe. If it encountered the edge of the wood at approximately a 5 degree angle or less, it was able to turn away enough to avoid falling. I was quite surprised by this, and thusly I think I've uncovered yet another trick behind the hexapod gait.

         

        You would think that when one leg goes off the edge, that side of the bug produces less thrust, and so the bug turns toward the edge and falls to its [possible] doom. But instead the rule my Hexbug seems to obey is exactly the opposite. Given a little thought, I realized why this is so:

        When one leg loses contact with the surface, the robot's balance becomes distributed amongst only 5 legs. In order to maintain balance, 3 legs must always have traction. With only 5 legs, the middle leg on the affected side ends up being on the ground for longer, because without adequate support, the chassis sinks down as the middle leg tries to raise up. It finally leaves the ground and the rear leg takes over, because the weight then shifts forward to the foreleg on the other side, which is supposed to be raised. So now there are two legs down on the other side when only one should be, and the middle leg on the affected side being on the ground longer than it should be. So the ultimate result is: the side with one less leg produces MORE thrust, and the other side produces LESS thrust, so the chassis actually turns AWAY from the edge!

         

        This is pretty fascinating to me, because with some development, a chassis could be designed that takes advantage of this quirk to passively implement edge avoidance, without any added complexity to the circuit.

        When I get a chance, I could build a 5 motor hexapod walker to test this further. The optimal implementation of this principle would probably be to make the forelegs the longest, with the mid legs being the shortest.

         

        Enjoy, Connor.


      • winglabs_inc
        Thanks! Likewise, I can analyze almost anything, but understanding mechanics to the point of analysis and building it are two very different things. ;) But
        Message 3 of 7 , Aug 10, 2014
        • 0 Attachment

          Thanks! Likewise, I can analyze almost anything, but understanding mechanics to the point of analysis and building it are two very different things. ;)

           

          But really, I have a shortcoming on mechanical know how, just the same. Which is why I sort of tend to simplify the mechanics and complicate the electronics. Seriously, on more than one occasion, I tried to build a walker by simply wrapping wire legs around the axles of motors, and using the motor(s) as the chassis. Needless to say, they ended up acting more as decorational pieces(spell check doesn't recognize decorational as a word?).

           

          But when it comes to Hexbugs, or anything arthropodomorphic, I like to analyze their behavior in great detail. In fact, Hexbugs are pretty much designed solely for this purpose. 

          Of coarse, it theoretically only works on hexapods. A "biped", tripod, or quadruped would do as you would expect and generate more thrust on the opposite side of the edge, pushing itself off. And an octopod would remain stable and continue forward until it fell off. There's just something special about that 3-3 gait that allows it to overcompensate and avoid the edge. That is, unlike an octopod or greater, the hexapod loses balance when one leg loses contact, but unlike a quadruped or lower, it's weight is able to redistribute itself amongst the other legs to regain that balance. Thusly, the gait is modified geometrically, without any change in pattern.

           

          And I'm sure that if the bot left tracks behind, they would seem to indicate that it deliberately tried to escape from falling, although it warily did no such thing.

           

          Also, I wonder if anyone else has noticed that the original Hexbug has an secret/unintended behavior? When it's battery gets low, it goes into a dormant state, only to be awakened by a touch to the feelers or a loud noise. It then reverse-turns and moves forward until it receives another stimulus, then reverse-turns and stops again. I think this is unintended, as it is not advertised as one of the Hexbug's functions, but ultimately a beneficial quirk. I've also noted, with dismay, that the Hexbug Crab is geometrically flawed, which must be why I rarely see them. That is, the Crab leans back as it reverse-turns, and this happens to be the side with the batteries, hence the heavier side. Mine very often ends up toppled over. Perhaps I could modify it's rear legs, to shift the center of balance closer to geometric center? In fact, I'm probably going to go do that now. After all, the Crab seems to possess behaviors that actually make a lot of sense for a BEAM bot, it's almost like a dumbed down version of Hider, and I'd like to observe it more closely to see if there are any quirks or emergent behaviors that I could take notes of, like with the Original.

           

          Enjoy, Connor.

        • tom_headoflead
          Hexbug, the little things they have at radio shack? I almost bought one the other day but I assumed they were too predictable. Do they resemble BEAM in any
          Message 4 of 7 , Aug 11, 2014
          • 0 Attachment
            Hexbug, the little things they have at radio shack? I almost bought one the other day but I assumed they were too predictable. Do they resemble BEAM in any way? I thought the scarab looked pretty cool.
          • winglabs_inc
            They re based directly off of BEAM robotics; the early ones can be considered BEAM(except for the RCs). They were just sold at RadioShack, but now they re
            Message 5 of 7 , Aug 12, 2014
            • 0 Attachment

              They're based directly off of BEAM robotics; the early ones can be considered BEAM(except for the RCs). They were just sold at RadioShack, but now they're pretty much everywhere. And some models are very predictable, yes. However, the Crab and Aquabot are a lot less predictable, changing their behavior at irregular intervals. And the Scarab and Nano are designed to behave chaotically.

               

              But in general, most BEAM outperforms Hexbugs, even a beetlebot or photopopper could display more immersive behavior than some models, and an average bicore or microcore walker could easily outperform any Hexbug model. In a sense, the early Hexbugs were not designed to operate in a real environment, but rather to demonstrate the principles of simple robotics and biomechanics to children, whereas BEAM generally takes this further and strives to build a device that can survive and operate for extended periods in a complex environment.

               

              Enjoy, Connor.

            • tom_headoflead
              Hi Conner,I did not know that. They are interesting looking, I was thinking how cool would it be to design a replacement PCB for them with actual BEAM
              Message 6 of 7 , Aug 12, 2014
              • 0 Attachment
                Hi Conner,I did not know that. They are interesting looking, I was thinking how cool would it be to design a replacement PCB for them with actual BEAM circuits? I may just have to get one for my son. The scarab looks to be the best one the inchworm is a neat design minus the remote control. -Tom
              • winglabs_inc
                I would suggest working with the Original if you want to mod it. But if you want it to be able to turn in more than 1 direction, you would have to completely
                Message 7 of 7 , Aug 14, 2014
                • 0 Attachment

                  I would suggest working with the Original if you want to mod it. But if you want it to be able to turn in more than 1 direction, you would have to completely gut it.

                  I have always been interested in the idea of a circuit that keeps the legs of a 2 motor hexapod in synch while turning, without mechanical aid. Basically, if you want your modded Hexbug to turn without messing up the gait, the number of rotations the bot turns for (x) must be a multiple of the difference in RPM between both sides, formally:

                   

                  {x = ([a - b])m}, where a and b are the motors, and m is the measure of the turn.

                   

                  Actually, I just realized why the Hexbug's normal circuit board looks so complicated for what it does; it doesn't just simply turn, it has to make sure that it follows a similar formula to what I described above. I suddenly have taken on a new appreciation for my Hexbug, as I've realized there is more calculation involved than is visually evident. But back to topic, you can still produce fluid behavior, if [a -b] is dynamic. You can also produce chaotic behavior by breaking or ignoring the rule. It would be useful to add pots to sense the position of each motor, so that a circuit can realign the legs, either automatically or manually.

                   

                  Enjoy, Connor.

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