PID Controlled Ping Pong Ball juggling Robot

A wooden plate juggles (and balances, while juggling) a ping pong ball (PID controlled).
Read more here: electrondust.com/2017/10/28/t...
things used on the plate:
-4 mics (for the determination of the ballposition)
-4 Savöx servos (for the movement)
-1 Mbed LPC1768 microcontroller (for the PID controlling)
-some analog signal processing circuits (for the mics)

[@galva1083]

LOL, I jumped to 0:39 and was wondering what kind of Alien tech is going on here!

[@danterusso6354]

Well done. Trajectory and proximity calculations have always been favorite algorithms of mine. I appreciate you sharing you work with us.

[@ElectronDust]

Thanks! Truth be told, this thing doesn't even use high level calculations. It just tries to figure out where the ball is every time it hits the table and then decides how much and in what direction to tilt the plate...

[@Maxisokol]

Hahah, this is a good continuation of ball/plate systems. ;)

[@kayipchan832]

A fantastic system!

[@jackc.2848]

0:39 "you are now entering the twilight zone"

[@The1wsx10]

you should set this up to make a stop-motion of a flying ball :D

[@ElectronDust]

cool idea! ^^

[@NK-fx1qs]

cool. where did you go with this? did you add some high velocity impact to really elevate the ball or anything cool?

[@Peter-sj6yn]

Hi Tobias I really like this. thanks for the video.

[@ElectronDust]

Hey Peter! Thanks. I really like your micro servo arm. Very cool! You inspired me to try to build a micro servo arm myself :)

[@julienmorris7051]

That s brilliant !

[@Robokong]

Love this project, going to have my robotics team build one. Do you have specs for the microphones and the controller you used?

[@ElectronDust]

Hey, Thanks! I used cheap electret mics. They are connected to an analog circuit which does the following: - amplify the signal - set a FlipFlop to High when the sound signal peals above a certain voltage level the micro controller (Mbed LPC1768) reads the FlipFlop states as fast as possible and counts how much difference there was in opposing mics getting set. btw, here is the same project with a camera: kzbin.info/www/bejne/b5atdGeIasyWsJI

[@peizhangzhu2656]

Could you share the sound frequency of the ping pong hitting the board? Thanks

[@raheels.1068]

Hi. Great work. Very nice project.

[@ElectronDust]

Hi. Thx!

[@user-eb8mp3ju1h]

I saw this video and made it similar using OpenCV. Thank you!!

[@ElectronDust]

Oh, I'd like to see it!

[@user-eb8mp3ju1h]

kzbin.info/www/bejne/gGevlGeDeZhlqNU here is the link

[@patriotir]

incredible!

[@trashz8015]

That's fantastic.

[@miguelangeljimenezhuerta1036]

Hi bro, I would like to build the project for school, it is amazing. Are the wooden bars that make the X shape fix ? Or they act like a scissor?

[@ElectronDust]

Hey Miguel! They are like scissors!

[@gortnewton4765]

So, here we are at 2020, 6000 years after the Pyramid of Cheops was built by Egyptians that we still can't understand how it was built and we have a table that bounces a ball... our civilization has advanced. Fantastic.

[@voyager5344]

Hi, thanks for sharing this awesome project. Do you think we can use STM32F103C8T6 instead of lpc 1768 on this project? Coz, this one is quite expensive comparing STM32. Thanks!

[@ElectronDust]

Microprocessor speed is an important factor since you are trying to get the time difference between the sound waves hitting the mics. looking at the datasheet of the STM32F103C8T6, it seems like it's max frequency is 72 MHz. If you do it the same way I did (processing the mic signals with a analog circuit and switch FlipFlops on as soon as the signal hits some threshold), then you can read the values (of the FlipFlops; one per mic) via digital input. So you at least do not have to worry about the speed of the ADC. But as said, speed matters. The LPC1768 runs at a max of 100 MHz. So you might argue that the difference probably isn't too big. Still, you might want to compare the datasheets carefully (and maybe take other microprocessors into consideration) before making a decision.

[@voyager5344]

Electron Dust thanks for your respond, I would like to ask what was the speed of the mics you used in this Project if you remember? I found mics available in speed of 20Hz-16KHz, I am not sure they are good to go, and i assume you used d type flip flops. I will try to work with stm32 microcontroller this project and let you know results, thanks!

[@ElectronDust]

Sorry for the late response. In the case of the mic, the 20Hz-16KHz Range isn't as much of a speed, but rather the bandwidth of soundwaves the mic can pick up. But you're probably aware of that. 20Hz-16KHz will do fine! I think I used D-Type FF's. Can't remember though. I just flipped them on and reset them all at once, once all 4 of them got triggered.

[@doomdjyo5997]

The machine is so good the ball floats 0:42

[@666PLATINO]

Hello dust, which servos did you used? I saw in the video’s description that u used vörtex servos but wich one?

[@ElectronDust]

Hey Martin, Those were Savöx SH-0262MG! Got to tell you though that the servos got a bit weak after a while. In the end, one of them wasn't able to lift the plate anymore. The weight was probably too much. Other servos might suit the job better! Need to be at least as fast as the Savöx. Speed matters.

[@666PLATINO]

Electron Dust Thank you so much

[@npoddubny]

Can you make 10 hour version please?

[@adityasidham1812]

Get some help

[@TutocarloCV]

nice

[@marcelorodriguez2321]

good very good project I wonder if could send information, schematics and codes

[@ElectronDust]

I think I posted a link to the code somewhere in the comments/description. There are no schematics. I just wait for a signal from the mics big enough to set some flip flop. While doing that I get the voltage levels of all the FFs (4, because there are 4 mics) and compare the time difference between opposing mics. The servos move around with PID.

[@hanakang2794]

whats the purpose of it

[@marcio6498]

Please, my friend, give us this opportunity and let us know the step-by-step of the project with the measurements and parts.

[@ElectronDust]

I have no measurements. And I already destroyed the thing. All you need is a flat plate as the initial position. In this position, the servos should be approximately rotated the same amount. Here's a blog post, just in case you didn't see it: electrondust.com/2017/10/28/the-quest-to-get-a-machine-to-juggle/ The important thing is to start. Just buy some servos and get them moving. after that, you can worry about the rest.

[@buzzmas8068]

But why?

[@wastetv9]

Its dam cool

[@ethanvx]

Hey that's pretty good

[@ElectronDust]

Hey Thanks! :)

[@marcio6498]

Hello my friend. Could you teach me how to build this project? It's for a school project. Please.

[@Mauropaz666]

what sensors are these?

[@ElectronDust]

Those are panasonic NaPiCa light sensors.

[@jeanlebosse]

It s genuis

[@Ryan-mg9fg]

Hey +Tobias Kuhn I'm doing something similar to this project but instead of bouncing the ball I am trying to keep it at a stable position like the following video kzbin.info/www/bejne/oGWynomCmMSietk but do you think you can explain to me what is calculating the balls position and where the machine should be going to? I know its using a PID loop or something but still a little unclear would love some insight!

[@ElectronDust]

Hey Ryan, I wrote an article about ping pong ball juggling here: robotshop.com/letsmakerobots/node/51318 Some more explanation: The timing for the upwards motion is fixed. This works, because the upwards motion speed is slow - fast - slow (because of the construction). If the ball hits the plate earlier than usual it will hit the fast moving plate. If it hits the plate later than usual, the plate will already have slowed down, so the energy transferred from the plate to the ball on impact is smaller than usual. Hope this helps.

[@ElectronDust]

Hey Rayn, I just noticed that I didn't really answer your questions. So let my try again: The balls position is calculated by using the sound it makes when it hit's the plate. And as you are suspecting, the thing is using a PID loop to calculate the next move. If the ball is quite far right, the machine tilts the plate to the left. if the ball is quite far right, but it is less far right then last time (the ball is moving towards the middle), then the machine tilts the plate to the left, but not as extreme as before.That's how the "PD" works. I actually didn't use the "I" part I think. In the ball balancing system the P and D parts are doing this: -P: tilt the plate in the opposite direction of the balls position (the center being in the middle of the plate) & tilt it harder the further the ball is away from the center. -D: tilt the plate in the opposite direction of the ball's movement. If the ball isn't moving at all, this value is 0. If it is moving to the right, this will cause a correction to the left. In the end you just add the P and D values and use it on the servos which are controlling the tilt on the plate. You might want to use one PID controller for the x axis and another one on the y axis.

[@abdenourouldturki7683]

could share the code ;)

[@ElectronDust]

Hey Nourri, Here you go: developer.mbed.org/users/pnpako/code/LCD_XpressoBoard_3_SoundSens/ It's a mess.

[@abdenourouldturki7683]

Well thank you some much and wich u devolopping more and more ;)

[@jiayuezhang4563]

666