A masterclass in over-engineering that applies rigorous technical discipline to a delightfully trivial pursuit. It proves that high-level expertise is most entertaining when solving problems that absolutely do not need to exist.
Deep Dive
Prerequisite Knowledge
- No data available.
Where to go next
- No data available.
Deep Dive
I Built a Robot to Raise 12 Tamagotchis
Added:I think this one is going to die, too.
I spent months building a robot that could take care of 12 Tamagotchis at the same time.
And for a moment, it actually looked like it worked.
But then one bug became 12 emergencies.
They're all dead.
I'm so sorry.
It's okay.
>> Before the robot could take over, I tried taking care of all 12 myself, and I immediately learned why this machine had to exist. This is a lot of work.
Shut up. I'm going as fast as I can. A Tamagotchi is a tiny '90s digital pet. I got mine when I was 8 years old. It lives on this little screen and all day long it needs food, games, medicine, disciplines, sleep, and cleaning up after. And you control all of that with just three buttons. If you do a good job, you get a very pretty tamagotchi, which is very healthy. And if you do a poor job, you get the ugly one with poor health. So, our robot must be able to see what's going on on the tiny little screen, and it must be able to press those three tiny little buttons. I had already built a robot for one Tamagotchi before. So, obviously, the next step was to make the same problem 12 times worse.
Mia, >> our first challenge is just keeping one Tamagotchi alive. And our final challenge is keeping 12 Tamagotchi alive for 48 hours. All original '90s Tamagotchis that I had were actually broken. Lucky for us, there's a modern remake. So, I bought every single listing from Finte. To get the robot to the Tamagotchis, my first idea was just to put all the Tamagotchis in a grid and drive the robot around. Then, the robot has to be really precise in where it stops. So, it's better to make the robot a station that the train of tomagotchis can pass through. I designed a nice and comfortable wagon for the Tamagotchi and we can actually link these wagons together to form the train. If this wheel stops even slightly in the wrong place, we will miss the button and a Tomagotchi will die. Therefore, I picked this mechanism for the rotation so we aligned perfectly with the buttons every time. And though this mechanism looks really sexy, it would become my enemy later on. Around this time, I started receiving my first involuntary test subject. To make sure that the whole train doesn't derail, I needed some bearings. And since I couldn't find the perfect product, I decided I'd do something I'd never done before. Make my own bearing.
Let's go, dude. Homemade bearings. To keep the Tamagotchis alive, we need to see what's happening on screen. So, we need a camera and we need a light. But the robot will get super confused with all these nasty reflections on the screen. So, we help the robot fight these reflections with a little tunnel.
And we use this ring light where we can disable any of the troublesome LEDs.
After wiring everything up, I found out that the slit we use to determine which wagon is number one actually isn't big enough. But later, I learned a ceiling light was preventing it from triggering.
But then the LED ring refused to work, and it turned out that the manufacturer had made a soldering mistake. Then I had made some wiring mistakes with the stepper motor, but a day later, I had fixed it all. By now, this wasn't just a sketchy prototype anymore. This machine had to run for days. This part of the video is sponsored by PCB Way. I used to find ordering custom PCBs kind of intimidating, but now I'm really glad I started doing it. On their site, you upload your files, you choose specs like board color, thickness, quantity, finish, and you place the order. And they also do 3D printing, which is very useful for all kinds of projects. So, with the electronics working, I brought my old code back to life on this new platform, and it can actually detect stuff on the screen. So, it turns out it's really hard to give this robot the right fingers to actually press the buttons. I made a whole bunch of different versions before finally I got it to run stably. And with that, our robot is able to take care of one Tammagotchi. You can see it cleaning poop for the Tamagotchi here. So, our next challenge is to keep two Tamagotchis alive with this machine. And I felt kind of bad while developing this robot because while I was doing that, these two were just getting neglected terribly. And I wasn't even close to the end of my abuse of these Tamagotchis.
For science, I wanted to find out how long it takes one to die if you don't care for it at all. 5 minutes we have the baby. About an hour later, the Tamagotchi has pooped itself and it's sick.
Yeah, the Tamagotchi didn't make it through the night. The healthy way to make a Tamagotchi happy is playing games. The fast way, snacks. This robot is not a good parent. It's an efficient one. After some more programming, I had the second one working as well. So, this means our next goal is to keep 12 of them alive.
To make sure no weird things like power dropouts happen, we give each Tamagotchi a fresh battery and then it's go time.
So, my niece has made it painfully obvious to me recently that dabbing isn't cool anymore. But now I'm cool again, right? Look at me.
It's happening.
Shut up. I'm going as fast as I can. I I mean, this looks ridiculous. It looks like I'm about to flash someone. Welcome to Tamagotchi Park.
Because of tiny differences in the Tamagotchis and the holders, we need to tell the robot exactly where each of the Tamagotchi's screens are. We actually reverse the camera distortion and then we get perfect pixel alignment. Then I built a script to test the buttons for each of the Tamagotchi. And we can view all the results in a web server. At this point, my life became fixing a vision issue, then a mechanical issue, then an electrical issue, and then doing it over and over again for weeks. The first issue is that the pillar with the camera and the servos is completely loose. Like this, the Tamagotchi have no chance. Fix it with longer screws.
>> Yeah, this is this is like really in place now. So, that's one problem down.
Another problem was that there was some movement of the Tamagotchi in the straps. So, I'm just putting some rubber band in there. And while we're developing and testing like this, I noticed something.
>> Tamagotchi is dying. This one is already dead. This one is dead.
>> Almost all of them are dead. Oh no. What a massacre.
I'm so sorry.
Can we save this one?
>> I don't know any Tamagotchi CPR.
>> It's okay to keep testing. We're just going to resurrect them. There's no escape here, Well, hello new friend.
>> Welcome to Tamagotchi Park. Anyway, back to the plan. So, now the buttons aren't working. So, once again, I replace the fingers. And even though the environment is quite harsh, life finds a way.
>> They're transforming.
Okay, if it now doesn't work, I'm just going to light this thing on fire with all the Tamagotchis attached to it.
But that actually did work now. But back to the schedule. We're getting these terrible white reflections because of the white clamps on the Tamagotchi.
Figured polarized filter might help. But since there's a lens in front of the Tamagotchi, that doesn't work. So you just get weird stains when you use a polarized filter. Then I tried replacing the reflections blocker by a smaller one, but it didn't help. The real fix was replacing all the white holders by black holders because they don't reflect that much light.
I refined the drive wheel a little bit because I have big plans with it. Now, the thing with complex mechatronic machines like this is they are really sensitive. So, you really have to handle them with care.
So, one time forward that works and backwards. Wow. part design was now tried and tested, but before I could trust it for 12 Tammagotchis, I wanted to upgrade it to aluminum. And conveniently, I had just gotten a CNC machine for exactly this kind of problem. But before I could do that, I first needed to learn how to not destroy circuit boards, plastic, plexiglass, and metals. Lucky for me, this machine is pretty smart with autoleveling and other calibrations, so it's pretty hard to destroy anything. Though, at first, I didn't really know how to install any tools, but it's super simple. All you just pull this thing and it just quick releases. So simple. And so in just a few hours, I had machined all the needed parts for this personalized test light.
And I felt ready to machine the big aluminum part later. Okay, after all this, it's time to do a bigger test.
We're going to try five Tamagotchi. And I'm just going to walk away and let it run for a while. Automatic. The best of luck, robot. So we'll be back in 1 hour.
When I came back, two were declared dead because the robot missed the skull icon on the screen. It thought that they needed food, but actually they needed medicine and unfortunately snacks are not healthcare. So, I fixed the code.
The Geneva drive also jammed, which is bad because when the wheel stops, parenting stops. So, I made the robot rehome every round just to remind it where the children are. Time for another test. It's about an hour later and look at this. All of them got detected properly and all of them are getting taken care of perfectly. The next challenge is to keep them alive for a full night.
I think it looks even better on the wall. I mean, just look at it. It's so cool. I also added the quiet night mode because the robot was mounted on the other side of my bedroom wall. It's a bit slower, but it's really quiet. Okay, this is it. The first night. Will the Tomagotchi survive? Will the robot actually do what it's supposed to do?
And will I sleep? That's the three big questions. Good luck.
Well, so far so good. All of the lights are off and the robot is just resting.
So, it's a brand new day and the robot actually did its job. This gives me the trust that we can handle a whole weekend. First night challenge is officially complete. The first night looked like a success, but at 2:30 a.m.
the camera disconnected and the robot froze. I thought it was a random crash.
Much later, I would learn it wasn't. So, filled with false confidence, I proceed into my preparations for the 48 hours challenge. So, this is where this awesome CNC becomes part of the robot.
>> 5 hours down, 2 and 1/2 hours to go.
>> It's fully enclosed, easy to work with, and very precise. So, I personally recommend this machine to makers and small businesses.
Now for some finishing touches. As much as I like this design and these colors, it reminds me too much of a different design.
Okay, that's better. So, from this point on, I can't touch anything or we will lose the challenge. It's all up to the robot now for the whole weekend. For about 5 minutes, everything looked good.
The robot revived all the Tamagotchis and moved from baby to baby, checking their hearts, giving food, pressing the buttons, and actually looking like a real caretaker. But then it missed one button press. And that tiny mistake puts the robot in an infinite loop. The 48hour challenge had failed after only 10 minutes. Somehow all the robot fingers are completely misaligned again.
Well, this really sucks because it's missing like a million button presses.
While this was supposed to be the big moment, I just want to go have dinner and leave them alone.
While I'm trying to repair all this, the tomagotchis are basically dying. So, I need to hurry up. But after I fix the fingers, actually, the vision is giving issues now because I probably bumped into the camera or moved the Tomagotchi around. Nothing is working for this tomagotchi. And it's pretty late already.
Dead. Dead. Dead. Also dead. This one.
Nope. Dead. Yeah, they're all dead and I feel pretty bad. But today we will make it all right. That morning I was greeted with a broken servo, but the birds were trying to cheer me up. Star Wars.
So, we're right back to our vicious circle. But I fixed the servo. I fixed the vision issues and it has now perfectly been running for 1 hour. But Friday, we learned that 1 hour is nothing. If it survives for 4 hours, I start the 48 hours run. Well, something weird is going on again. You see these markings? Let's fix it.
And we secure that drive to the axis.
The machine has been running stably for 4 hours. So, it's finally time to start the 48 hours challenge. Let's go.
It's the end of day two and the machine is doing perfectly.
So, it's now the second morning and all the Tamagotchi are still children, but I have a feeling that they will become teenagers very soon. Okay, we did it.
The robot has been running for 48 hours and all the Tamagotchis are still alive.
This is what I've been working towards for months. It actually did the job. And now, as a bonus, I'm just going to keep it running and see what happened. Some of the children have transformed into teenagers. Here's one. This is the bad caretaking one. I think if you make too many care mistakes, this happens. And here's another one. And then this is the good caretaking one.
>> This is night number three. Now the road is still working perfectly.
All of the Tamagotchi are teenagers now and there's three ugly ones and there's like nine pretty ones.
It's day five and something really bad has happened. I came in to check on the robot and it says here that one of them has died. I couldn't really believe it because the robot has been working perfectly.
>> It actually died.
>> Yay. Our robot produced an adult. Well, then when I came home on day six, the robot had frozen. All of them needed urgent care, but even worse, three of them had died. No.
>> At this point, I finally learned what was the cause of all the freezes. I was slowly filling up the memory with junk until the program couldn't run anymore.
So, the robot hasn't stopped for a moment, but apparently, we lost even more Tamagotchis, and now we're just left with three adults. But later that day, we lost another two. Lucky number three, still alive. But then came day number nine. The last Tamagotchi just died after 8 days, thousands of button presses, a lot of Tamagotchi dead, and a slowly dying CPU. The daycare is finally empty. But the goal never was to make immortal Tamagotchi. The goal was to find out if one robot could take care of 12 of them. And the answer is yes. So after 8 days, what did the robot actually achieve? Well, all these Tamagotchis got to live rich and full lives. And it went not perfectly, not super responsibly, but enough to prove that this machine actually works. And now I'm pretty happy to have my quiet workspace back. Want to support more ridiculous builds like this? Click join to become a channel member. You'll get custom emoji, loyalty badges, and your name in the supporter list. See you next time. And you can watch the first Tamagotchi robot right here.
Related Videos

Setting up a curved screen with Immersive Calibration Pro 4 and multiple cameras (P3D v4)
FlyerOneZero
23K views•2019-07-21

Robot Learning with Sparsity and Scarcity
allenai
379 views•2025-10-14

Jorge Mendez-Mendez: Unlocking Lifelong Robot Learning With Modularity (2023-10-05)
umassmlfl
237 views•2024-01-06

Northwestern’s MS in Robotics: Student Robotics Projects, 2023
NorthwesternEngineering
1K views•2024-05-31

"Perfect" Turns: Turning by the Gyro - FIRST LEGO League (FLL) SPIKE Prime + EV3 RePlay Programming
ZacharyTrautwein
94K views•2020-10-02

Gorkem Secer: TSLIP-based Deadbeat Running Control of Bipedal Robot ATRIAS
DynamicWalking-wv6qm
298 views•2018-06-22

Self-Driving Cars Need Lessons On Human Drivers | Maddie About Science
skunkbear
26K views•2018-08-21

Milrem Robotics’ THeMIS UGVs used in a live-fire manned-unmanned teaming exercise
MilremRobotics
99K views•2021-05-20
Trending

we're almost finished the house (ep.125)
JennaPhipps
347K views•2026-07-22

We Finally Know Where Saturn’s Rings Came From
astrumspace
79K views•2026-07-22

BIG BET: Cathie Wood goes ALL IN on Elon Musk
FoxBusiness
89K views•2026-07-22

MIC DROP: Smithsonian Director Called Out For Woke Propaganda
TheAmalaEkpunobi
37K views•2026-07-23