Close
0%
0%

Quadruped spider robot

building a quadruped robot with sg90 micro servos

Public Chat
Similar projects worth following
Looks like I have to update this description.
I won't go after an opendog clone anymore, because it would be even more difficult and physics is torturing me anyways.
Now I opted for a four legged spider robot with sg90 micro servos.
Right now just using an arduino nano, because they are easy and fit onto the robot.

Please don't read this. It needs an update. Thank you :)

The journey upto this point was a pure horrorshow. Everything cracks. Everything breaks multiple times. Nothing really works. It got really really frustrating at times.

My plan was to get this robot with me to 35c3, but I got so mad just before that that I deleted everything and just started from scratch after I did calm down. Gladly I had a working prototype which i just had to remodel in openSCAD.


The equations I use to control my legs currently come from these two links (apperently both use exactly the same screenshot, so I don't know whom I have to give the credit):

https://www.faizbalweel.com/single-post/2016/03/18/Inverse-Kinematics-3DOF-Quadruped

https://oscarliang.com/inverse-kinematics-and-trigonometry-basics/


Gotta draw up all the schematics and all of that for my exact robot. Hopefully I can convince myself doing that since uni is almost over for this semester and I have time now.

  • 1 × Arduino nano (or maybe esp8266)
  • 12 × sg90 micro servos
  • 1 × pca9685 pwm controller
  • 1 × 2 cell lipo batterry

  • It's back alive

    Mario04/13/2019 at 21:45 0 comments

    After months of doing nothing I got it back to live. It shrunk a bit, I switched over to the esp32, but it's alive.

    Prototype creep walk works. Gotta wrap my head around how to turn itself the best and write up on all of this stuff. The openSCAD model needs a rework, but that won't matter for now and will probably one of the last things to come.


  • F.

    Mario01/20/2019 at 12:07 0 comments

    Another servos killed itself today and I have no spare servos left.  Maybe this is a good point to take a pause.

    It was a really good learning experience though. Quite some lessons learned:

    - Don't overestimate tiny plastic servos. They can barely take anything.

    - Physics will make you run in a lot of traps.

    - Robots are fucking shit, but a lot of fun, too.

    - Hacking, to see quick results, is okay, but approaching every step quite slowly will probably protect you from a lot of frustration.

    I think it is time to move on to a bigger robot, but first some other fun stuff.

    PS: All openSCAD files and my current arduino code are in its repositories.

  • Wiring hell and other controller

    Mario01/19/2019 at 22:53 0 comments

    I tried to tame the wiring hell. To achieve that I got rid of everything that is not needed right now (LEDs, LCD). Also switched to the esp8266 again for easy of remote control. Possible will add the LCD back again later, for a status display (Network working etc.). Hopefully I can get myself to upgrade the repositories soon.

  • inverse Kinematics

    Mario01/18/2019 at 15:12 0 comments


    As I have already said I have the inverse kinematics from this website:

    https://www.faizbalweel.com/single-post/2016/03/18/Inverse-Kinematics-3DOF-Quadruped

    I'll just wrap it all up with a few pics now.

    Suppose we want to move our leg to position (x, y, z):

    First we have to calculate the angle which our leg has to turn (see picture below). Note here I mounted the leg on the servo that way that the 90° position of the servo is 45° off the actual wanted coordinate system. So make sure you'll just add pi/4 at the end.


    From that we'll just have a simple equation:

    Now we have 1 of 3 servos finished. The other two are not much harder either. The picture show's the angles that we want to calculate. We also got the following constants:

    C => The length of the coxa

    F => The length of the Femur

    T => The length of the Tibia

    Therefore we get the following equations:

    Therefore L is:


    which then lets us easily calculate alpha



    Now only beta is left which can be calculated like this:

    Those are all computations needed to achieve inverse kinematics for this exact robot.

    Be warned though. You might have to rotate your coordinate systems to get this working, because the servos are obviously connected in a mirrored order. I might publish a log post for this sometime soon (implying).


  • Moar LEDs

    Mario01/13/2019 at 19:26 0 comments

    Needs more LEDs

View all 5 project logs

Enjoy this project?

Share

Discussions

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates