Let's Make Robots!

m&m sorter #3

Sorts m&m's by color
DeltaRobot-0.02.zip3.44 KB

Videos: http://www.youtube.com/watch?v=Pbb9GOWaL4g and http://www.youtube.com/watch?v=mPakG8p3MHE

Update: Added schematic for the vacuum system

Code word were: CHEAP!

Cheapest RC servos I could get, Cheapest Arduino, almost. Straws from mc donalds, icecream sticks, rubber bands, Lots of scrapped stuff just waiting to be used in the workshop. Solenoids in the valves were from old tapedrives, and the valves them selfs were from old bicycle tubes. Vacuum and pressure comes from an old aquarium pump, reshaped to make vacuum too.

The dark disk is a sun shade, had to add that before filming outside, othervise color detection made too many mistakes.

"Exploded view" of lower arm:

"Exploded view" of lower arm

Lower arm:

Lower arm

Lower arm about to be connected to bolt on upper arm

Lower arm about to be connected to bolt on upper arm

Lower arm and the moving platform:

Lower arm and the moving platform


Two valves, the lower one disassembled

The pin visible at the end of the yellow tube pin of the bicycle valve. The brass tobe above goes over the pin. Since the pressure uses in the robot is low, +/- 0.15 bar, I could go with these simple home made solenoid valves.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Well, the main component in the vacuum detector were the small vacuum chamber which would compact a little when vacuum were applied to it. Then a small screw which were adjusted to the point where it touches the vacuum chamber under normal pressure conditions.

With both vacuum chamber and screw where made of metal it were easy to detect if they touched each other, simply measure the electric conductivity. When the joint stopped conducting, the vacuum had reached the threshold and the program would new that i could continue with next state.

Dear Mogul,

I have design and implemented my Delta Robot.

But when I move it to move positions then the moving plate is not parallel to the base plate. It has some tilt in the moving plate.

Can you tell me how I can remain the moving plate always parallel to the base plate ?

Is there any special condition to make the moving plate always parallel to the moving plate ?



Strange, it should be. The end effector, the "plate" that's moves around, should be dead parallel with the base. It should be given my mechanics only. It sounds like you have not made the lower arms as twin arms. Look at the photos again, pay attention to the lower arms, those made of straws, there are six of them, 3 pairs. They are always parallel in pairs.

I think the delta robot is actually so damn cool because of the extreme simplicity in the design, and it's really easy to control as well. I have built many handling robots, but the delta has proven much easier than anything I have tried before.

May I suggest you to enter your robot here on LMR, and add some photos of your creation, then maybe I, or somebody else here, can give you some advises on how to fix your construction.

Show us what you got and lets get that darn thing moving!

Dear Mogul,

Your robot is really coo.

I have one question. Can you tell me the reference point of the motor for your Inverse kinematics. I have read some articles about the inverse Kinematics but no one clearly indicate the Zero Angle of the Motor or Zero Angle Position.

Please tell me the zero angle position of the motor.



I did not use any reference points other than placing the upper arms horizontal when the servos were in middle position (90 degrees)

Absolutely amazing! Does it work with Skittles or Gummy Bears, too? ;D


Your robot is amazing mogul :) I really like that, good job. I decide to creat my own :). But I am just beginner. If you do not mind, maybe you could post some good information, website or books, where I can find and learn how to build it. I will be very, very grateful.


Sorry Sir, but that not how it works. Honest, I do not want to sound rude, but I think you should start making your own thoughts about how you want it.

Try something, anything is fine, but go on, try! Make an all mechanic model demonstrating that you are able to get the movements going, just use your hands as motors.

Or start at the other end, grab a micro controller and 3 servos, see if you can get them to do "anything"

Have you ever written a program for such a thing? If not, I suggest you start with a simpler project. The "Start Here Robot" is designed to absolute beginners, you can even buy a kit.

I'm not trying to scare you or anything, and trust me, we are a whole gang of LMR'ers here willing to help you with YOUR project. Go on start on something and feel free to ask when ever you get stuck.

Now, back to my robot. I watched a lot of videos on youtube showing delta robots in action. Then I read some theoretical articles on the geometry, but that were too advanced for my math skills. Then I toyed with different cheap materials and settled on wood, straws, glue, rubber bands.

Later on I made the sucker system with pumps valves and stuff. If you have specific questions about any details please ask!

Set out a goal you think you can reach, start moving, when you are closing in on your goal, set out a new goal and repeat.

And let me repeat, I'm trying my very best to help you learning how to build your own robots. And, one more thing, remember to have a good time while doing it!

The robot is fantastic. I have 2 questions (1) with more resources/funding could you build a more "sturdy" version (2) I noticed that the M&Ms were covered after being placed in the vision detector is this because the ambient light affected your color read?

Thanks for your kind words.

Yes, I'm quite sure I could build it more sturdy, stable and precise if my goal were not to do it as cheap as possible. Go search LMR for "delta" and most of the hits will lead you to more solid designs. Mine is very primitive compared to what others have done. (my trade is not craftsmanship)

Yes, the black lid is to cancel out ambient light. Before I "invented" that, my home made color detector had an very high error rate, 30 to 40 percent, if I remember correctly. Now with the lid, it have not made a single mistake, and trust me, it has made several thousand detections.