• Mats Önnerby posted an update 7 months ago

    I made a small upgrade of the shoulder servo. The small potentiometer broke, so I use a larger potentiometer with a 6mm shaft. It’s the olny part that I have changed,but I think the same principle can be used on all potentiometers. The new potentioneter can rotate 270 degrees instead of the original 180 degrees, so the servo mapping has to be…[Read more]

    • Hello Mats!
      That for sure is a safe pot!!
      But you didn’t use my latest pot holders, that means your parts can no longer fit PivTit and RotTit of the latest version…
      I guess, it’s because you already had PivTit and RotTit printed from the previous version. I like your little potentiometer board option, it’s handy for to simply disconnect.

  • I’m almost finished with the mechanical build and the testing of each servo individally. Next step is to purchase / build and install all the electronics and batteries. The back is still a bit empty.

    • Super cool !!
      Your choice of colors remind me of Leonardo’ super hero:

      http://inmoov.fr/members/leonardo/album/

      How is the Raspi doing for post processing video tracking on MyRobotLab?
      I will post the Back Parts soon…

      • You should post your pictures in your album, you might get featured on the side page.

      • I get about 2 FPS when doing facetracking. That is good enough, however I also get 5 seconds latency, and that a bit to much. I have tried using the camera outside MRL and I get good results from the camera. I can stream at about 30 fps about .2 s latency. That is good. I had a problem that took me some time to figure out. The java runtime that…[Read more]

        • I would love to have this on my desk, I only have a head so far, and that’s controlled by LEGO

        • I also realised some time ago that using the latest version of MRL which uses OpenCV 3.0 that I would get a low fps result. I was recently mentionning that to Grog and he said he would digg into it.
          So, using the Lenovo tablet touchpad with the version 1.0.107 gets me with a a fps of approximately 30.
          Using the version 1.0.954, gets me to a 7/10 fps.

  • The new dual track servo pulleys interfere with the cover of the hand. If you use them some material needs to be removed from the arm cover to make room for the front middle pulley. Thanks to Juerg who warned me that this could be a problem.

    • Hi Mats,

      Is this a option to use:

      https://copy.com/1K7hNQFiuvHMBHUh

      (sorry, there is no option to upload a picture directly in a commend)
      As you can see at the drawing, there is a normal sized part (left side) and a smaller sized part (right side).
      For the smaller side (with a oval form) you have to play to get the correct length you need.
      If…[Read more]

    • Hi Marten. I just cut of a few mm’s of the hand cover support. It’s only there to support the structure during print. I don’t want to replace the dual track pulleys since they work very well. Have a look here for a demo https://www.youtube.com/watch?v=b5S-8hjsMr8

      • Hi Mats,

        I’ve seen the video and the results are perfect !
        The option was more meant to update the Pulley’s for other InMoov builders, how like to use your Pulley’s, so they haven’t to cut anything ;-)

        • From a math point of view, an excentric pulley will work very different. With two tracks, it’s just linear movements. The excentric can only work between very limited angles. With two tracks, you can rotate any angle. The ratio will still be the same.

    • Maybe reducing the filet height for the tendon, this way the pulley would be less high and wouldn’t cause a problem with the covers of the for arm.

    • Awesome solution for tensors Mats!
      Thanks a lot for it and for sharing it over thingverse!

      Saved me $$ on springs!

      :D

      • Sorry forgot to ask if this solution need a different MRL configuration, if so can you be so kind to explain changes need please?
        Thanks again!

  • I added a new video to show the difference between the original servo pulley for the hand, and the new dual track pulley. https://www.youtube.com/watch?v=b5S-8hjsMr8

  • A litte update to expain the reason that I redesigned the servo pulleys. I use the fingerstarter to explain. In the first picture the hand is open and both lines are stretched. On the second picture the hand is closed, but the lines are not stretched any more. To close the finger the servo pulls about 27mm and releases 27mm on the other end. But…[Read more]

    • Pictures were uploaded in a different order than I expected. So the last picture shows the open finger with lines stretched and the two other when it’s closed and the lines are not stretched anymore.

      • hi mats
        Maybe I see things the wrong way but I do not understand why you have to modify the pully. Set your servo to the middle position than add the pully with a 10 mm offset towards the closing side. In MRL define a mapping that covers the 44 mm range for the full movement with your servo?

  • Video of testing the new servo pulleys for the hand. https://www.youtube.com/watch?v=gwx3JIakDV0

  • I have made new servo pulleys for the hand. They have two tracks with different diameters, since you need to pull a longer distance to close the hand than to open it.http://www.thingiverse.com/thing:1074604

    • To close one finger you need to pull 27mm but to open it again only 17mm, The 10mm in difference will result in that the lines easily fall out of the tracks, The servos will also have a deadband of 10mm before they start to pull, With this construction the lines will be stretched all the time, so the fingers will keep their position better.

    • Good question. I don’t think that springs are necessary, but if you turn the wrist, the lines will be tightened. So you have to compensate for that by relaxing the tension on the lines a little,

      • Hi Mats
        Can’t really see why you want to modify the pully. Put your servo to middle position and add the pully with an offset to the closing position that will account for the different distances for opening and closing. The use the mapping feature of mrl to adjust the overall distance for your servo in use?

      • I posted an update to explain the probem.

  • First test of the servos. The neck works fine. Jaws needs to be changed to increase the speed of movement. Eye servos tested but not connected to the eyes.

    • That looks like a very good design I really like the eye movement

    • Hey cool ! Thanks for posting the video!! I love to see stuff like that. Yes the speed of the jaw needs to be increased. Bob Houston has made a remix for the eyes, I wonder if it fits in one normal InMoov head.

  • The eye mechanism can be found here: http://www.thingiverse.com/thing:1035688

  • I’m almost finished with the head. Next step is to connect the servos to the Arduino and power supply.

  • I had several ideas when I started to print the head. The large servo was more expensive than a small, so I tried to replace it with a standard size sero. I also wanted to mount a better camera in the head than a USB camera. My modified head contains a Raspberry PI 2 with a Raspberry Pi Camera module mounted in one of the eyes. It also contains…[Read more]

    • wooouu, setting the brain in the skull.
      What sort of camera did you set in?
      I tested my full InMoov with the Raspi2 and was a bit disapointed with latency. The Odroid U3 was also having latency…
      Let us know your Rasp2 acts when you will be using speech recognition along with OpenCV tracking, so we can share comparison.

      • I use a Raspberry PI Camera module connected directly to the Raspberry CSI port. I has really good quality for still pictures and video stream is OK. I tried to use it with MRL and the OpenCV service, but I got some error messages. I will sort it out with the developers of MRL. Thay are all very nice, skilled and helpful people.

        • Others have tested the Raspi Camera module into MRL with success. Your errors might be related to which version of MRL you are using.
          When I tested the camera with the Raspi2, I was using version 1.1.107 and it was working fine, though with latency.

  • This is a part of the Bicep. I use PLA, and it's not as soft as ABS. So when I tried to mount them together, it was impossible, because it wouldn't flex. But PLA gets soft at relativley low temperatures, so I used a heatgun at a low setting to heat the connections. After heating the plastics for a short while, it was easy to mount them together.…[Read more]

  • Hand and forarm printed in black PLA. I use recessed M3 screws to hold it together. I drill a 3mm hole in the part to be mounted. In the part that it should be mounted to, I first drill a 2.3 mm hole and then I use a 3mm taps to make a thread. It's not easy to make the thread straigth by hand, so I mount it in a drill.The 8mm screws are made from…[Read more]

    • Tabing the holes really helps, it is what I also do. Unfortunately not everyone has this kind of tools.
      Great work by the way!