Robot Cloning by DIY 3d printers!

The DARwIn-OP is a open hardware and software project which greatly aided me in doing this project.

        DARwIn-OP is an acronym for (Dynamic Anthropomorphic Robot with Intelligence - Open Platform) or DARwIn for short. The DARwIn-OP was developed by the RoMeLa research lab at Virginia Tech in collaboration with University of Pennsylvania, Purdue University and ROBOTIS. ROBOTIS is a world leading South Korean robotics’ company. Their Dynamixel servos are the leading robotic servos in the world and the key to why the DARwIn-OP is so ground breaking. The lead designer of the DARwIn-OP project has Dr. Dennis Hong of Virginia Tech’s RoMeLa and ROBOTIS. The robot is a state of the art research and development humanoid robot. The DARwIn-OP weighs in at about 2.9 kilograms and a height of 45.5 cm.

The idea behind this project is to find new and cheaper ways to create and build robots. As 3d printing becomes cheaper and easier to use more and more people will have access to this way of manufacturing. I hope that this instrucable will  open up more peoples eyes as to the power of this new personal fabrication method.

Below is a link to my Make magazine article on how I made the clone.

http://blog.makezine.com/2013/04/26/cloning-the-darwin-op/

Step 1: Why I wanted to own a DARwIn-OP!

I thought it would be a good idea to show a production version of the DARwIn-OP in action to show the reason why I wanted this cool robot.

This is video of the DARwIn-OP in its natural environment Robocup soccer.

 

Step 2: Download CAD file

First step was to download all of the 3d CAD files.

Link to file location.

Other download information links

Step 3: Create .STL files for all of the parts.

I used Autodesk Inventor to create my .STL files, you can use your favorite 3d design software.

If you need a cheaper way try AutoDesk 123D Beta if you can still download it.

Here is a link to my Thingiverse project page for my DARwIn-OP clone where I have upload all of the .STL files for the robot.

http://www.thingiverse.com/thing:9793

Step 4: Print out all parts on a 3d printer.

I used the UP! Plus 3d printer to print out almost all of the robot, again you can use a 3d printer that you own or have access to. This took me almost two months to get all of the parts printed out correctly. 

UP! Plus 3d printer.

The Afinia H-Series: 3D Printer is the exact same printer.

Step 5: Buy the electronics for the robot.

Link to website.

The easiest part of this project.

Step 6: Buy your servos for the robot.

Yes the MX-28T servos are expensive but they are state of the art robotic servos. The servos of your robot are alwas the key as to how well your robot will preform.

Link to where to buy them.

Step 7: Buy fastners for the robot.

It may not seem like it but this was one of the hardest parts of this projected. I was aided by the detail list in the assembly manual. After a lot of phone calls and e-mails and internet research I found McMaster-Carr in Chicago was the best place to order fasteners from.

http://www.mcmaster.com/

Step 8: Assembly Robot

The DARwIn-OP has three very detail manuals that you can download. They are an Assembly, Wiring and Fabrication manuals. 

Step 9: Download software to robot and laptop

The control software is open source and works on windows and Linux.

DARwIn-OP support website.

Step 10: Turning on robot for the first time

This is always a very scary moment when doing this for the first time.

http://youtu.be/z-RA5v5_5i4

Step 11: Test the robot

This is also a terrifying moment but it can also bring great joy!

http://youtu.be/5VuZTewMSKY

Step 12: Redesign parts that break

This brings out the great advantage of personal fabrication. The ability to make part after part until you get it right or iteration.

Step 13: Print out new and improved parts

How cool is it to have a 3d printer of your own!

Step 14: Reassembly robot with the new parts.

I like building robots so I enjoy this part!

Step 15: Retest robot with new parts.

This can be boring but it has to be done.

 

http://youtu.be/z-RA5v5_5i4

Step 16: Loop

Do While ( robot Fails )

{
     CALL step 12,
     CALL step 13,
     CALL step 14,
     CALL step 15,
}

Step 17: Robot is working!

If you are like me, you will we always be going back to step 14 because of new ideas or ways to make your robot better. 

Total cost for my project was 6000 USD if you don't include the 3d printers that I used. A lot of money yes but if  you buy a factory made DARwIn-OP from Robotis the cost is 12,000.00 USD. So I saved about 50% of the cost of the robot by building it myself.

If you are interested in following future upgrades and mods to this robot you can check out my blog.

http://mike-ibioloid.blogspot.com/

Step 18: Future upgrade possibilities.

Modifying the covers of the DARwIn can lead to some interesting modification ideas. 

Simplest would be changing there color to hardest which would be modifying their design.

A great example is changing the color of the covers to mimic the Ironman power suit.

Jet-pack and weapons for the power suit may take some more time and research.

The images where created in AutoDesk Inventor by my friend Yoshihiro Shibata.

Comments

[moverstreet007]

This is another video of a DARwIn shows off its soccer skills, the programming was by a South Korean team.

[moverstreet007]

You can also check out Make Magazine volume #34, page 66 for more details on how I made the clone.

http://blog.makezine.com/2013/04/26/cloning-the-darwin-op/

[moverstreet007]

You can also check out Make Magazine volume #34, page 66 for more details on how I made the clone.

http://blog.makezine.com/2013/04/26/cloning-the-darwin-op/

[shojiki]

Yes thanks for your reply and I hope I can get started soon!

[shojiki]

That was cool! Anyway I am also a programmer and am always interested in robotic but didn't know how to get started and How do you learn to design these cool robots? Are you a self-learner?
Thanks,
Vic

[moverstreet007]

Yes, I am a self-Learner. I would tell you jump in and get started.

[Oz]

The original DARwIn-OP, which is sold by Robotis, has metal parts. For strructural integrity I always thought that one cannot build the whole robot using a 3D printer.
Which parts of your robot failed first? Were those the metal replacement ones? Is your version lighter than Robotis version? Did that affect the kinetics?

[moverstreet007]

Yes the DARwIn-OP brackets are made from aluminum.

I think people just assumed that you could not print out the brackets. I think I am opening a lot of eyes to the possibilities of what 3d printing can do.

The UP printer uses ABS that is very similar to ABSplus the Dimension printers use.

My version is lighter but you can can changes parameters in the DARwIn software as it is open source.

I have found that 3d printed parts are more flexible than metal and that strength is not that big of a issue.

[drewnoakes]

I too assumed that you couldn't print the aluminium structure. You've definitely opened my eyes with this! Where are you based? If you were able to visit Germany for the RoboCup German Open in April, or Holland for the World Cup in June, I think a *lot* of people would be really fascinated to see this for themselves, myself included!

[moverstreet007]

I wish that I could! That would be a expensive trip for me.

I live Kansas City, Missouri, USA

I should be going to the Bay Area, Kansas City, Detroit and New York maker faires this year and Robogames.

[lpergola]

is this an open platform? can i get darwin schematics to modify it ?

[drewnoakes]

All the details needed for the structure, skin, electronics and software are freely available on SourceForge. There are component sheets, assembly manuals and CAD files aplenty.

[moverstreet007]

Yes, the DARwin-OP is a open hardware and software project. I have links in the Instructable as to where to find and download them. If you need more information, I will try to find it.

[TedRobotBuilder]

Nice! Idea: couldn't you put the Darwin electronics in one of their cheaper models, like the BIOLOID Premium Kit, and end up with a robot very similar to the Darwin but only about $2500? Maybe spend more and start with the GP?

[moverstreet007]

Ted, many people think that the Bioloid GP was a early prototype of the DARwin-OP.

You can also find early examples of the DARwIn-OP that used AX-18s too.

That is my next project idea. A cheaper DARwIn-OP that uses AX-18 servos, a raspberry pi and the CM-900 servo controller. Just need to find time in the day to do it!!:(

[TedRobotBuilder]

The Darwin board is basically an R-Pi isn't it! Nice way to slash $1000 right there! XD

The downside to these bots though is they can't carry much in the way of sensors though. We need to shrink the Kinects sensor down to its size! :D

[moverstreet007]

The DARwIn has a FitPc2 for a brain and a CM-700 servo controller.

http://www.fit-pc.com/web/

http://www.robotis.com/xe/darwin_en

But that is a good idea though, you can use the raspberry pi and the CM-900.

check here for how to do it.
http://www.robotsource.org/xe/index.php?mid=Circle_CM9_Developer_World&page=2&document_srl=10624

[moverstreet007]

Sorry CM-730

[moverstreet007]

I forgot, I think the Kinect has already been shrunk. This is the Kinect 2

http://spectrum.ieee.org/automaton/robotics/robotics-hardware/handson-with-the-next-generation-kinect-primesense-capri#disqus_thread

I need to redesign the head but it would fit!

[drewnoakes]

Not to be a bummer, but if you wanted to compete in the RoboCup Humanoid Soccer competition with this guy, you wouldn't be allowed to use a Kinect, or any active sensor for that matter.

Also, I don't think a Raspberry Pi would really cut it in terms of processing power to make a viable robot. Even the FitPc2 is sometimes lacking in speed with its two (hyperthreaded) cores. Probably the most demanding requirement of the robot is vision processing. The software that Robotis provides does a very basic job of colour segmentation, but it's not robust enough for soccer competitions. I'd really love to see a small GPGPU on the platform somehow.

[moverstreet007]

I should have been more clear. I was thinking about the Auton. Humanoid Challenges at Robogames.

http://robogames.net/events.php

Not sure if I will ever have the money for 4 DARwIns and the money to go to a Robocup event and compete in it.

[drewnoakes]

Cheers for the link. I hadn't heard of that event before. Fira is another popular robotics event. I'm working with a university in the UK on a new team of DARwIn's, and am very grateful for the opportunity. We've submitted our qualification video and are waiting to see if we will compete in June. It's an amazing platform, and the fact that you made one yourself is fascinating. I really wonder how different it would be in practice. You mentioned that it's lighter and some parts have broken. Great that you can remodel those parts. Thanks a lot for sharing!

[moverstreet007]

yes for robocup:

"
Humanoid

Soccer Humanoid League

In the Humanoid League, autonomous robots with a human-like body plan and human-like senses play soccer against each other. Dynamic walking, running, and kicking the ball while maintaining balance, visual perception of the ball, other players, and the field, self-localization, and team play are among the many research issues investigated in the league.
The league is divided in 3 subleagues, according to robot sizes: Teen Size, Kid Size and Adult Size."

[moverstreet007]

Funny DARwIn-OP video to watch:

[sarveshk]

They are not servos... they are Dynamixel. Dynamixel is a patented product designed for rapid development of automation in robots. It can be used in place of servos but I think good quality hobby servos as sufficient for this robot. Is there any specific reason for which you have used Dynamixel in place of servos?

[moverstreet007]

Sarveshk, i would agree that Dynamixel servos are more expensive then regular servos.

My question to you is what servos would you use?

The DARwIn-OP was design as a research and development robot.

A great example of what it is used for is Robocup. I don't think you would get very far in Robocup using standard servos. What do you think?

[sarveshk]

I watched your video. This result is difficult to attain with a hobby servo. So everything is fair in love and a [competition] ( where you have enough funding!).
Go for Dynamixel then!
Still I think, for regular hobby use and low-cost research, I think servo is sufficient assuming you are able to tweak a servo for feedback.

[syates3]

I'm really impressed with the Darwin design I like how responsive it is and streamlined vs the Japanese version power means nothing if you can't move fast enough to use it.

[Tomdf]

Gasp, those are some pricey servos! How do they perform compared to regular servos? Faster and stronger? What do you think would happen if I tried cheap servos instead?

[TedRobotBuilder]

These servos report their position. Normal RC servos don't. You need that for robotics.

[moverstreet007]

Yes to every you said TedRobotBuilder:)

[sarveshk]

There are plenty of methods by which you can read the servo position. That might not be as accurate as a Dynamixel, but sufficient for many hobby applications.

One method is to use a magnetic sensor. A guy had made a nice magnetic sensor and that was very cheap. It gave great results and he was selling it a few years back. I would post the link of that if I find it.
Another obvious method is make hole in the servo and read the pot directly. Here's a tutorial for it. http://letsmakerobots.com/node/27206

Of course in any of these methods you would have to open and modify the servo that might void your warranty. But looking at the price difference between a Dynamixel and Servo, warranty is not a big issue.

[TedRobotBuilder]

They make much cheaper Dynamixels, these are just top of the line. They also give you load, temp, greater range of movement and other things. You could rig this up too but it would be a lot of work for all the servers. Load is very important for walking.

[moverstreet007]

I don't want to discourage you from trying but!

When most people talk about regular servos they mean 10.00 analog servos from China. I have used them on lots of projects but they just do not have the holding torque, speed or resolution for humanoid robots.

http://www.robotis.com/xe/dynamixel_en

Even the low end AX-12s from ROBOTIS blow them away.

The only servo line that compares too them is the Kondo line of high end servos.

http://www.kondo-robot.com/EN/wp/?cat=14

I would tell you good luck and keep me updated on your progress.

[Tomdf]

I see. After reading the specs I started to get it. Full 360 encoding isn't something a normal servo can do.

[moverstreet007]

Tomdf, MX-28T are butt kicking servos. High speed, high holding torque, almost no backlash during movement. They are almost perfect servos for humanoid robotic uses.

[drewnoakes]

The MX-28T have metal gears and built-in PID controllers as well. You can read the load on them, the temperature and voltage levels too. http://support.robotis.com/en/techsupport_eng.htm#product/dynamixel/rx_series/mx-28.htm

[moverstreet007]

Yes and with an onboard 32bit 72mhz Cortex M3,

yes very sweet servos!!

[chaitanyak]

wow! thanks for sharing

[moverstreet007]

Thanks!