The wireless quadruped made from $10 in parts and materials.

Public Chat
Similar projects worth following
This walker is a exercise in minimalism. Can I make robot for $10 that can be easily assembled with no adhesive? Can I make the instructions good enough that it can be assembled by middle school students and work every time? I want it to be elegant and the assembly to be a satisfying experience where every part just feels right. This should be a great starter project for soldering, Arduino, and digital fabrication at a maker space. This is completely open source and the files are all included.

My goals for this project are as follows:

$10 BOM cost, batteries not included.

No glue used in assembly

The parts should be made in 30 minutes or less

Parts should be easily shippable

Assembly should be manageable for a middle school student

Every piece should fit together with a sense that it belongs

Ideally I teach my daughter how to run a business from this project

Make it repeatable version of the very one off 

Tortellini the Turtle bot I made 8 years ago

Now it is time to do at least  as good I as I did with this bot in a reproducible way. The above bot was all glued together with 4 motors and 4 photo sensors. The MCU was 16F pic and it ran on an interrupt driven code but  it was definitely a one off labor of love. There was zero design for manufacturing incorporated in the design. 

Now that I belong to the makerspace I have access to a laser cutter which makes very reproducible cuts. My initial goal was to make product it with acrylic, some off the shelf hardware and the other electronic bits and bobs from Aliexpress.  Now I am thinking about adding a few 3D printed piece if it would significantly improve the performance.

I have worked on it for over 2 years, made over 50 different revisions of the laser program but there was always something not quite finished. I decided to start to publish and rev as I go along. Once it looks solid, I can put it on Tindie and make that $1 I have in my yearly goals. I have bought the hardware to support a Tindie run of 50 pcs many times so I have boxes of hardware in the basement.

MPEG-4 Video - 25.36 MB - 02/06/2019 at 04:43


JPEG Image - 1.27 MB - 02/06/2019 at 04:41


MPEG-4 Video - 47.09 MB - 02/06/2019 at 04:23


Standard Tesselated Geometry - 87.19 kB - 01/22/2019 at 02:59


Standard Tesselated Geometry - 325.08 kB - 01/22/2019 at 02:59


View all 12 files

View all 14 components

  • Reving up for Version 17 - A fix list

    shane.snipe06/17/2019 at 02:03 0 comments

    Well our laser cutter at the Fox.Build makerspace in St. Charles went sideways for a while and I did not have confidence that I could make good parts. It seems to be back on line so I am going to take another swing at this. The items I had concerns about are:

    1) Too much friction. Mostly because the gears are too wobbly. The 3D printed axles worked pretty well but the pin to hold them in place was never done right. 

    --- To reduce part count, I will look at making a 3D printed axle which collapses on it self to get through the hole and then splays out to keep it in place on the other side. It may take some cut and try as my old Japanese boss used to say but I could save 8 parts and make assembly as snap, (Pun intended.)

    2) More could be done to lock the board in place.

    3) I am considering the worth of using better drivers. The L293D are the cheapest drivers on the planet at 0.25 but I think I am losing a volt. I might pay $0.75 extra to have a snappy robot. Maybe the mechanics can make it efficient enough that I will not need it but it is on the radar.

    4) I want to implement a BMS. I am always concerned about running my expensive batteries into the ground. If you deplete the Li ion batteries too much, they will stop working or possible explode when charged. 

  • Version 16 Prebuild Fixes

    shane.snipe02/16/2019 at 15:26 0 comments


    Tested the photo diode sensors on Version 15 and was not seeing a high enough voltage  for the ESP8266 to get the high input. Decided to run the the resistor before the sensor instead of after. Breadboarding it showed promise and so I redid the board. I also had the wrong pitch on the motor connectors so I changed it from 1.25 pitch to 1.5 mm pitch. OSH Park set a notice that they are shipping so I need to get the rest of the changes implemented.

    Acrylic laser cut parts

    Our laser seems to be losing a little steam. Unfortunately everything was dialed in for 10mm/s and 80%. I need to find the new sweet spot. The  areas of poor fit from V15 that I will fix and release a new dxf are as follows:

    1) Shaft fit for the motor gear and space was too tight. Add a 0.1 negative sew compensation to see if it helps.

    2) Make the board holder z height line to line instead of the 0.3mm clearance.

    Lastly, I still have not figured out how to manage the laser cut files well. I like to modify in Fusion 360 because the push pull feature is cool, even if there is no design history. However, they can not output a full assembly drawing to a DXF. It can output an individual sketch in DXF but I want to keep the relative spacing for he assembly.

    So the workflow is to make the adjustments in Fusion and then expect the Step file. Import it to Onshape, make a drawing and then export the DXF.  Here is the link.


    The tatoo grommets I used for feet are awesome. I tried to replace the drive gear clip with them and it seams to be working very well. This will be one of the main none electronics parts.

    or Allexpress

    The new axle/gear holders are working great. We just reverted our Prusa MK2.5S to direct drive so I hope the holes do not get filed in like last time. The shape is perfect though so I hope I the print better. 

  • Build Report Version 15

    shane.snipe02/04/2019 at 05:50 0 comments

    Overall it has been a success. I tried Kicad for the first time and the board is largely functional. The new 3D printed fastener concept is working well and the gears are held in better than they have ever been.. But as I mentioned before, it is all in the tenths of a mm and here are things that need to be changed for version 16.


    I have gotten the 1.25 pitch instead of 1.5 pitch for the connector. It kind of works but I have to bend the pins to force it.  I will fix it today.

    However, the big success was the 7805/2 Li ion battery combination. The motors now have just the right speed. It is awesome.

    It looks like i may need to put the resistor before the photodiode instead of after.  This will allow a higher voltage when the photodiode provides a high signal.

    Acrylic Frame

    Our laser cutter seems to be losing some steam. I cleaned the lens but could no longer cut at 10 mm/s. This means all of the dialed in parts get a little wonky.

    1) Shafts were a little tight for the motor gear and spacer. Need to open it up by a tenth.

    2) Encoder is a little loose on the shaft. Tighten it by a tenth.

    3) Motor fit in the wall was awesome. 

    4)  Board holder was too loose in the vertical direction. I had 0.3mm slop. Combined with 0.2 mm 

    kerf it was sitting on a 10-20 degree angle and I had to glue it.  Will go back to line to line. 


    This was my second round with the 3d printed gear axle/holders. 

    The play in the axial direction is perfect.

    The window in the 3D printed parts to fit in the clip is getting covered over in the print and I need to file it. I works pretty good after filing but that is not idea. Maybe I will make the window bigger in the radial direction by 0.4mm.

    The shaft clips are breaking. There is not enough flex. Increase the radius at the root of the U by 0.2mm. Replaced it with a foot (tatoo grommet) and it seems to work well. Might be a permanent fix.

    Here's to round 16!

  • Battle of the tenths

    shane.snipe01/22/2019 at 02:32 0 comments

    Can you believe tenth of a mm can make a huge difference.  It will make the difference between a loose fit and that satisfying perfect press. 

    So I tried out my new fasteners and they worked better than my creations usually do without tweaking. The holes were to small because 3D printed holes shrink more than I ever compensate for so I will just make them bigger. To test the pieces I used some 1.8 instead of 2.7mm acrylic and the gears spun really well. They could be a half mm more snug to the face.

    The battery holding scheme is awesome. One plastic rivet and a fence. Perfectly satisfying design. 

    Here is a list of tweaks for this next run:

    1) 0.15 mm offset for the clips to hold on the gears.

    2) 0.1mm wider ledge for the T-Clips to hold the legs in place.

    3) 1mm bigger holes in the 3D printed axles.

    4) Added the straps to the dxf. 

    5) Rounded the sharp corners on the walls for fun.

    I also started to get the Fusion 360 part tree. To move or copy a component you need to highlight it on the tree not the screen. Make a sketch, then a component and copy the component for multiple copies that can be edited from one sketch and moved independently. I was making multiple sketches for multiple parts. I meant lots of work if I wanted to change something. 

    One more item was that I found a dxf converter so at least i do not have to go back to onshape.

    Here is the latest update. I will put it in the files as Version 15.1

  • Version 15 Laser Trial

    shane.snipe01/20/2019 at 17:49 0 comments

    I have completed my CAD to do list.  I like the way I can work in Fusion but I have not found a good work flow to get the parts into a sheet form for laser cutting. I ended up exporting a DXF of each part, saving them all in one sketch. Extruding them all. Making a drawing. Saving the dwg. Not being able to save the dxf. Export the step file of the extruded drawing. Importing it into ONSHAPE. Making a drawing and saving the DXF. THERE HAS GOT TO BE A BETTER WAY!!!!  In Solidworks I had an assembly of the laser cut layout. If  I changed something in the model, it would change in the laser cut assembly and I would just output a dxf. The drawing would update automatically.

    In any case, here are the changes I made. It is a big rev so I hope it works.

    1) Changed from proprietary hardware to 3D printed axles and laser cut spring clips.  I printed out the parts but my printer was not having the best of days. We will change the Prusa back to direct drive and I think it will do better. I may need to open up the holes. I had to do a fair bit of filing to get in the spring clips.

    2) Added holes for plastic rivets to hold down the batteries.

    3) Change the board holder. Thanks to Kicad's mechanical output, I think it is in the perfect place. 

    Today  I will laser cut the 2.7mm acrylic I buy at Menards and see if I can get some nice friction free walking.  I have the new Li Ion batteries. (10440 x2) and the new board that will work with the higher voltage from these batteries is being fabbed at OSHPark.  

    Here is what the layout looks like. Some how it is much sharper in real life.

  • Fasteners

    shane.snipe01/15/2019 at 04:25 0 comments

    I had planned to use Chicago screws or book binding rivets. However, not only are they an additional cost and a impediment for someone to reproduce my work, the were binding when it was walking. The Chicago screws would come unscrewed and crash into the leg. Although I liked the feature that the robot could be disassembled, the has got to be a better way. 

    So after a few days of head scratching, I came up with a 3D printed and laser cut hybrid. First of all, this is more square pegs in round holes to try to reduce the points of friction. This robot does not go fast enough to worry about friction welding but the motors do not have enough juice to overcome a lot of friction.

    The second feature is the super low profile and simple to print shape. There is a bit of bridging but I think the Prusa will print it like a charm. I went line to line on the vertical walls and gave it and extra 0.3mm on the overhang length. Lets see if my estimation is good. 

    The other great thing is that this removes 4 of the 8 counterbores from the laser. It will speed up the cut by a few minutes because it takes a long time to scan. I hand added the counterbores to adjust to a specific rivet width.

    Here are some pictures.

  • Fusion and Non-Native Geometry

    shane.snipe01/13/2019 at 15:11 0 comments

    I am new to Fusion 360. I started working with it 45 days ago. However, I am blown away buy the ability to work with non-native files like step files. On Solidworks or Onshape, at Step file is just a piece of material. It does not recognize any features so you do not have much flexibility.

    With Fusion, you can click on a hole and delete it. You can click on surface and type Q for push pull and you can move it around. Thanks to this flexibility, I already have my PCB mounted and the pcb mount reworked. Thanks to KiCad spitting out the the file, I know exactly where it is rather about where it might be. Here is the top view. 

    And from the back.

    From the bottom.

  • CAD to do list

    shane.snipe01/13/2019 at 12:16 0 comments

    I have moved from Solidworks to Onshape and now to Fusion 360.  I am struggling greatly with the pin and slot joint in Fusion even though I got it to work in Onshape and Solidworks.  It works sometimes with contact surfaces but then it breaks. In any case, here are updates need to do for version 15. 

    1) I added 2 10440 Li ion batteries. One on each end. Some fastening method for the battery holder needs to be developed.

    2) I am not happy with the Chicago screws that hold on the gears and I want to design a 3D printed alternative. It will take $0.75 out and make the walking smoother if I do it right. I am thinking about a type of friction weld to hold the 3D printed rivets in place for the gear axles.

    3) Solidworks has the concept of parts and I could change the parts and the separate assembly that is all ready for laser cutting changed as well. Fusion is not set up like this. I read that people make a separate components for laser cutting using the same sketches. That way they update automatically. This makes sense and I want to try it.

    4) I need to hold the board in place better. Now that I have the 7805 dropping below the board  have something to hold on to so I may be able to hold it in place so the encoders actually work.

  • ECAD to Fusion 360

    shane.snipe01/13/2019 at 11:30 0 comments

    I have always kept my PCBs and electrical parts separate from my models. With KiCAD I learned how to output the 3D board with any components that were in the system. The output is an easily editable step file and I can add the component that were not in the library in Fusion 360. Here is a picture of my new board. 

  • Battery Power 2

    shane.snipe01/12/2019 at 12:00 0 comments

    Today  ordered 4 AAA Li Ion batteries and a charger from Amazon.

    I also bought some single battery holders.

    By distributing the weight front and back, the balance will improve and the robot will not fall over while taking a step.

    By using the 2 Li Ion I will get 2 benefits.  Reduced weight over 4 regular or NiMH batteries and the combined higher voltage will give me more speed for my motors.

    The down side is I will have to modify my circuit to add a 7805 voltage regulator. Maybe I can use the 7805 part as a battery stop hanging below my board.

    In my circuit, to make the motors go faster, more than 5V is helpful. 7-8 V might be idea. However, the Wemos can only handle around 6 V before letting out its magic smoke. I have a diode in the circuit now to reduce the voltage by 0.6V but I think I will replace the diode with the 7805 voltage regulator to allow the 2 Li ion batteries to be used at their full charge of 8.4V to their depletion charge of about 6V.

View all 13 project logs

Enjoy this project?



Dejan Ristic wrote 02/07/2019 at 08:36 point

Very cool project! Looking forward to future developments. 

I saw in the video that Tortellini could pivot left while at a standstill, but can it also make a turn while walking forward? That's one of the main issues I'm wrestling with in my own project.

Would love to see some more info on the leg movement scheme you're using. And more videos. (No pressure though, just a suggestion.)

  Are you sure? yes | no

wheaty wrote 01/22/2019 at 04:58 point

Great Job! I Think this project has a lot of potential.  I like that you are focusing on low BoM cost, ease of manufacture, and ease of construction. 

  Are you sure? yes | no

Pat Helmers wrote 01/18/2019 at 20:38 point

I can see this being a great project for my middle school nephews. A lot of possibility for applied robotics with practical applications. I'm following closely.

  Are you sure? yes | no

shane.snipe wrote 01/13/2019 at 15:30 point

Does anyone know how to work Fusion 360 joints well. I really would like to make a video of my simulated walking gait but I can not get the pin and slot to work. I made it work in Onshape and Solidworks but Fusion wants the slot to be in a specific direction I think. In any case, I can not get the gear to move the leg and it slide in the appropriate slot. Here is the link in Fusion if any one wants to take a shot at making it move.

  Are you sure? yes | no

Mike Szczys wrote 01/10/2019 at 16:39 point

Nice! The movement in the video is really entertaining (very turtle-like ;-D ). Can we see some pictures of the hardware without the shell?

  Are you sure? yes | no

shane.snipe wrote 01/11/2019 at 03:16 point

I will see if I can dig it up.

  Are you sure? yes | no

deʃhipu wrote 01/09/2019 at 17:22 point

That is so adorable! Great job!

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

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