Hexapod Makeover

[blegas78]

I have always had a soft spot in my heart for hexapods since the first time I saw one around eight years ago. Unfortunately I have never had the excuse to spent the great deal of time and money on a good hexapod until this last semester, where I built a reinforcement learning hexapod, where it learns to walk based on vision. Being crunched for time and money, I could only use mostly what I had in my garage, and I had to program in a very simple programming language, Matlab. When the successful project finished, I was left with a job in a robotics lab, and a possible research project for my dissertation when I go into graduate school. Realizing that I will use my hexapod for the next robotics class and for future research, I now have a great desire to indulge myself in sophisticated hardware.

I have begun porting all of the code form Matlab to C++. This is nearly complete, and the most interesting part of the code is the simulation built using OpenCV. I have posted the video in the other thread, but here it is again:




I have just finished ordering the last set of motors. The set up will include six RX-10 motors for the coxas, six RX-28 motors for the femurs, and six AX-12 motors for the tibias. I have three extra AX-12 motors which I may use for the 3 DOF head/camera.

As I mentioned in the other thread, the lab I work in has a Dimension 3D printer, which is essentially like having a print button in Solidworks. The possibilities are nearly endless with this machine, save for a few constraints. Models are printed layer by layer with 0.17mm ABS plastic with .01mm resolution (I think anyway). The machine has two cartridges, one for model material and the other support material. The support material is used for support for overhanging extruding features or holes. This way, any sort of complex geometry can be built, geometry much more complex than what a CNC mill can build. The only problem is that it can be difficult or impossible to remove the support material in situations, for example if there are many complexly routed tubes inside of a block. There are solutions for this, such as the better version of the 3D printer in the lab, which has a chemical soluble support material, but I am too happy with the available printer to complain.

I will do my best to post some before and after pictures. Firstly, here are some shots of the old hexapod:




I downloaded Solidworks last week so I'm still a little inexperienced with it. Here is what I plan to print, and please, any suggestions or comments are greatly welcome!

Body top plate (bottom plate almost identical):


Coxa:


Femur:


Tibia bracket (I will use old thread rod and brass tube):


Here is the final assembly:

[blegas78]

So far I have printed the top plate and a tibia bracket. They both took a couple tries because I am apparently bad at measuring and Solidworks, but now everything fits snugly.

Here is an awesome quality image of the 3D printer taken with my phone. Unfortunately I could not find the lab camera today. You can see the bottom plate, two coxa brackets, and four tibia brackets completed, with two femurs just about finished. The printing head is the blurry part with the yellow triangle on it.

[plingboot]

I've just had a look at the 'uPrint' demo film on the Dimension site and can't really believe that device is for real (!!)

[Matt Denton]

Nice work! I edited your original post so the youtube link works.

The 3D printer is a peach! want one!!!

Would be interested to see your OpenCV code..

[blegas78]

Here's my new motor arsenal for the legs:

[blegas78]

Nice work! I edited your original post so the youtube link works.

The 3D printer is a peach! want one!!!

Would be interested to see your OpenCV code..

Thanks for the fix, I'll be sure to do it that way from now on.

Are you interested in only the simulation part?

[blegas78]

The old and new femur are comparable in weight, but the 3D printed versions are much stronger. Here they are together:

[Zenta]

Hi,

Great project!
I really envy you the 3D printer. What kind of material is it? It seem to be very solid. I didn't know you could make part like that being strong enough. Wow!

And I really envy you the nice collection of Dynamixel servos!

Keep up the great work!

-Zenta

[blegas78]

Hi,

Great project!
I really envy you the 3D printer. What kind of material is it? It seem to be very solid. I didn't know you could make part like that being strong enough. Wow!

And I really envy you the nice collection of Dynamixel servos!

Keep up the great work!

-Zenta

Thanks!

The material is ABS plastic. it is not as strong as normal ABS, since the process makes the model a little porous. It is a little flexible, so if I made the femurs one sided like what is typically done, then the flexibility of the plastic would make it a very poor material. Fortunately, the dynamixel motors have a mount on the other side of the horn in the same axis of rotation, so I can make these "H" femurs and they become incredibly rigid when mounted. The femurs compared to my older design are a little less flexible than the brass version, which is great.

I have assembled the body, here's a before (a little frankenstein-ish):


And after:




I can't get over the fit, everything is perfect. I'm not used to having things be so perfect! Also, this body is lighter and MUCH more rigid than the old version. I don't notice any flex when I apply torsional forces to it.

[wuselfuzz]

incredible.