Neural Network Robot With Arduino

[@nikodembartnik]

Support my projects on Patreon → www.patreon.com/nikodembartnik

[@NitishYadav-lb7zc]

Please make more content of that kind 😊😊😊😊😊😊 THATS AMAZING at least for me

[@TechBots_yt]

Can you see my project one time🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻😥😥😥😥😓

[@shawnekuriakose]

where is the github code

[@invonatormoustaphaalihssai2114]

Pleas expriment code

[@vrankyrule]

How much ram does this have??

[@shaur1221]

And here I am just trying my best to make a blinking led

[@Kaelidian]

😂😂

[@ishuxgaming8187]

😂😂iam trying to just connect wire

[@patricialim6816]

Same 😂

[@RDKCREATIONS]

🖕🖕🖕🖕🤭🤭🤣🤣🤣😂😂😂🤣🤣😂🤣🤣

[@shaur1221]

@@RDKCREATIONS :0

[@potbot887]

In case you want to extend this project, I have few suggestions: Use Webots simulator to simulate your robot, that way you can train the neural net and see the output on your computer itself and try out various environment tests, before running it on actual robot itself. second, if your neural network works differently the next time you train it, use fixed set of initial weights, typically done by using "random seed" or just save the initial weights in a file and load it again the next time you train.

[@milankarakas3005]

I am glad you got it working. I just made similar thing, but only manual algorithm. Instead ultrasonic, I am using ToF (Time of Flight) optical sensor VL53L0X which is good between 30 mm and 1200 mm. First it moves servo left and right at 5 positions and measures distances, then chose the longest distance and goes there. After approaching obstacle, it goes slightly back and then repeat the operation of finding more space for drive.

[@inferno6012]

the first test did not work because it was scared first time in the world you know

[@OlivierCR]

Makes sense

[@nikodembartnik]

Yeah, we should give it a little bit of time to let it adapt to the humans

[@dbs777229f]

I love to see the fails, it makes me feel better about my issues.

[@Christian-mn8dh]

Steve Washick it makes it more realistic

[@realVertiqo]

Wanted to say that a couple times already but always forgot in the end. Your Cardboard Explanations are top notch Dude! Very very well made and easy to understand for younger folks especially i think.

[@nikodembartnik]

Thanks, those take a lot of work to make them, but outcome is amazing!

[@ValleyMustang]

Can somebody show me what you guys mean by these cardboard explanations

[@realVertiqo]

@@ValleyMustang The way he explains stuff with cardboard cut outs and scribbles on it style. for example minute 0:50 onwards.

[@farhanamin4075]

Bro do you know where is schematics

[@owenoj]

I've built basically the same thing, the only difference is, is I let the sensor take a reading from left and right then position the sensor forward, which is its default position and determine which direction to move towards where the object is farthest away. When it detects it's getting close, it stops, looks left and right and does the same thing again. Therefore the sensor is always looking forward whilst moving, and only looking left and right when it's stopped which is when it's gotten too close to something.

[@danimaster6647]

Hey Nikodem, since you were speaking about something with more computing power and mentioned the RasPi: There is the Corel Dev Board available. This has a tensor processing unit and this makes it very powerful for neural networks. It runs models about 25 times faster then on a good CPU and costs about 150 Euro. This might be worth checking out for you, you can even do object detection/classification and other CNN stuff. Best regards

[@flingmonkey5494]

I have had trouble with the ultrasonic sensors, I sometimes get bogus short readings from them, due to the wide detection cone and echos from the floor itself. I started experimenting with taking three readings and accepting the highest value I got. This smoothed things out. Also, I suggest you make the robot move a little slower, allowing more time to take the readings. One challenge I would love to take would be to take the basic car you have there, rip off the ultrasonic sensor, throw in a gyroscope, and make it accept commands like "turn to x degrees" and "forward", and so on. These commands would be coming from a second Arduino that controls all the sensors (SoftwareSerial communications.) Once you have a base driveable vehicle that can drive a straight line and turn to any heading, construct another Arduino or RPi system that senses the surroundings and issues commands to the drive system.

[@bassome3000ify]

Really cool and awesome 👍👌💪 1-train the neural network on your computer, then copy its weights to Arduino (train in computer, implement in Arduino) 2-check neurona library, on the developer's website there's a browser based playground

[@drakethomas652]

Just a suggestion for the sensor, I think you should try out a lidar or laser sensor, since at a certain angle, the ultrasonic sensor tends to send back false data.

[@digihz_data]

Or use 3 static sensors instead of 1 on a servo :)

[@romeshfernando2959]

Use three ultrasonic sensors because you can get three inputs(distances) at the same time. As an example , if it takes one second to get distance of one side, in your robot it will take three seconds for three sides. but when you use three sensors it will take one second for all three sides. I have ddeveloped one and I used three sensors + back sensor . It is working fine

[@Pmc07AyeUrDa]

Very good. The working montages added extra quality to the video.

[@TheTinkerDad]

Ok, this is really cool. Never thought something like a neural network is possible on a low power device like an Arduino.

[@frankai.3713]

Running a trained net is not a big deal

[@maciekm7953]

Ciekawy projekt 👍 siedzi się godzinami i walczy z problemami ale jak w końcu zadziała to jaka radość 😁

[@omarcruz6326]

That's VERY COOL !!! 😎 Great job, Sir.

[@gggalpi]

"Could be done with an if statement", " the problem is that the robot is on" hahaha very funny, nice content

[@parthjoshi5779]

5:58 to 6:22 perfect MacBook air advertisement maybe

[@mathisbuilder]

Bad computer lol

[@code-grammardude5974]

I am currently making a Raspberry pi robot and I have a webcam on it. It is convenient because you can use great Python libraries like Tensorflow and Keras to make Neural Networks.

[@joetke]

I've posted a comment without reading yours. Tensorflow + Keras wrapped in python = terrific combo

[@supermodernotv]

The nice thing of this approach is the you get to learn how those NNs work starting from scratch. All thise frameworks are great but they abrstract a lot of what is going on underneath. This project is great fro learning the basics a bit like when you blink an LED to learn the basics of microcontrollers

[@navanda4676]

Here is and idea. If you make a game that are a virtual representation of the "car". Then you can an random generated "labyrinth" and put a neural network in multiple cars and use an Genetic Algorithm instead. That makes it faster to lean and you don't have to monitory it. Plus you make the genetic algorithm a powerful computer instead off the arduino. Like a said just an i idea, really liked it video. Keep it up!

[@sarabhian2270]

so finally I got something to do 👀, what I will do now is , I have same ultrasonic sensor and something similar moving bot, i will try something new ,thanks buddy 👍

[@supermodernotv]

Good job, keep making cool videos

[@nikodembartnik]

WOW! Thanks Massimo! :D

[@joetke]

When I first saw the title vs. the video length I was perplex: talking about neural networks in 10 minutes... On top of it you use arduino.. hummm. That should be a big challenge I thought. Then you made a pretty good demonstration of a good methodology, embedding that library and eventually seeking a better one which led you to an almost success. Congrats! I'm happy as well because your background matches mine regarding serious coding backed by some maths. Looks like Poland takes advantage of its past ties with the USSR and their scientific education which relies heavily on maths. (Obama said that american students lack that solid basis). Anyway, We've implemented on a RaspberryPi 3B a machine learning using google tensorflow (there are decent tutorials about it on youtube for those in a hurry, and google provides straight-forward applications too). It works flawlessly and we intend to build a robot like yours. I do think that it's pretty hard to dive into details when it comes to explain neural networks to a large audience. It could be really boring for some people ;) Your animation may help Last word: 10 minutes to explain what you've really done to make your robot respond properly, is really, really too short. It may frustrate some of us. It's a pity. But I'm looking forward to watching better documented AI applications at a slower pace ;) Thank you Nikodem!

[@nikodembartnik]

I try to keep my videos rather short and entertaining :) I know sometimes it's hard to explain stuff really well in 10 minutes, but if you want perfect explanation of NN there is already a lot of great videos on youtube about that. I also don't know enough about them to make such video :)

[@joetke]

@@nikodembartnik you're a honest guy.

[@entiretwix1480]

I had a dream about this, then today it was in my recommended

[@DevineIshaq107]

Don't need AI for such operations, 8 direction functions, 3 range finders with comparison operators: the bot will pave through obstacles and can be trained fo pass through a gate

[@Pedritox0953]

Niko this is an excellent project... the trainning in NN is very sloooow in general... also i recomend to use supervised learning... what is that? Simulate the input and output of the NN to train the weights... another recommendation is that for sensors fix is better due to noise... place them like 30 degrees... simulate input and output as desired... train the NN... Enjoy!! Cheers

[@muhamadekanugraha6882]

That robot was like my life without a purpouse, and being told just don't make a mistake.

[@goodn1051]

Am a simple man... I see NN,I click,I subscribe and I like

[@NachiketaHebbar]

Great Job, Loved the experiment! The only reason for it not working with great accuracy could be a small dataset, which is the backbone of a neural network. Increase the data set by simply providing left, right , center values and which direction the bot should move ,in a csv file. However that seems unnecessary here, as the problem can be solved with simple if else statements . Nevertheless , nice video. Best of luck !

[@NitishYadav-lb7zc]

Please create more robots of that kind 😊😊😊😊

[@BuildEver]

Dlaczego uczysz sieć korzystając z "wolnego" Arduino? Moim zdaniem taką sieć powinieneś wytrenować za pomocą swojego zdecydowanie szybszego komputera, a potem przenieść model sieci do arduino. Przykładem może być narzędzie Tensorflow Lite Arduino. Środowisko do obsługi takiej sieci, może dać średnie wyniki ze względu na małą ilość pamięci Arduino Uno, oraz wolny procesor, ale będą one zdecydowanie lepsze niż teraz. A przynajmniej szybciej zobaczysz efekt. Edit -- Teraz zobaczyłem, że to filmik z przed roku

[@sajidahmed5079]

Test out side of home...you r getting reflection from wall...so robot is rotating in all direction..or set range of detection of sensor

[@tthtlc]

Good video, now I am motivated to do something.....

[@bitnbytes01]

Awesome video but my spidy senses were detecting misaligned picture frames in the background but I like that :), chaos is needed in this line of work. I will definitely try this AI robot kit out. Good project.

[@muhammadbasir83]

I thought it was only me who put a piece of tape on the webcam. haha

[@rodlupanow69]

Just great, congrats. I want to try it too.

[@go_better]

Very interesting episode! Thanks!

[@TanishqIsHere]

A Pi-Zero would be a much better option for this, have a look at Sipeed MAIX too.

[@atanaspeltekov]

The problem is not in the arduino he is enough powerful the problem is in the code

[@DanyMaster]

Of course this experiment could be better if you use other platform, but you've used Arduino (understanding this limitations) I think the result is very cool... Well done mate!!

[@chathurafernando1299]

Great work mate but you should minimize the speed of motors and servo. And next time use the " interrupt ". It's more easier.

[@gordonyoud5975]

Slow down the sensor swing, if it is looking for a return signal after sending out a pulse, the echo might be missed. :-))

[@enos5192]

No one : Me Looking at the head : Why didn't anyone fit a radar in there??

[@Jdozjsnwj783]

I think you might not understand neural networks... If you do then your're not sure how to translate this to arduino, but with a simple training set the bot should be able to recognize certain distances and stop effectively, I think your code may be limited by the inaccuracy of the HRS0-04 ultrasound sensors, they're not really that reliable

[@NeoPh03nix-lk8fc]

hi

[@mohamedfarid7499]

AI and ML will add advantage to our arduino projects to decrease codes time and effort

[@SuperADI2]

If you are really passionate and curious in this thing s i recommend you what I do, an university in Robotics and Artificial Intelligence followed by many years of programming and developing gadgets, some are made public on my KZbin channel some are not and i will see if I will have time to upload it in future

[@kiszkilukiszki4057]

You are from poland?

[@EngineerFromPakistan]

This project can be completed with just if conditions. how neural networks are making this project different?

[@ffoska]

Write the node weights to eeprom. Pretty cool project !

[@nikodembartnik]

Thanks! Yes, that would be a good idea :)

[@valentinburillier9029]

Good job, it's an interesting video. In order for your robot to learn you will have to integrate Q-learninge and not base you on a dataset. Your robot will learn with a system of reward

[@ernknn]

I really like your content. Can you tell me what kind of battery you used for the car? Also, what kind of DC bench power supply are you using? Thanks!

[@mateuszspec1658]

Po akcencie poznałem że to Polska XD Pozdrawiam, mega robocik :D

[@advanceautomatedsystemsgmb9311]

At last you did Copy and Paste 😁

[@erumanbredberry9238]

Молодчина, Никодим )

[@joseiltonricardo]

Você é muito novo, com toda certeza terá muito tempo para implementar suas ideias em redes neurais.. Parabéns pelo vídeo!!!

[@arthuraraujo8190]

Just use 3 sensors, so it would not have any delay

[@smipy]

8:33 - 8:39 Robot : Go Stupid... Go Crazy.....

[@SanthoshKumar-ud4ui]

I think this doesn't need neural networks,just reading input from sensor and set of if loops.

[@elmerperez7350]

it doesn't, but who cares? NN are awesome

[@imiebaka]

Keep it up lad, Nice video ...

[@mikegLXIVMM]

Great work!

[@lucsonnelson8037]

Great video brother. Only thing is I wish that I could see hir you implemented the sensors and the motors in the sketch. Please I kinda want to build something similar with more sensors, but don't know how to link the NN in my project. Help 🙏🙏🙏

[@farahsumudlayon7890]

you're so good!

[@DIVERGEsound]

Nice video!

[@oELxTOROo]

It would be cool to see if you can get a simple machine learning program on a raspberry pi!

[@primodernious]

the commodore neural net linear matriz is a good ezample of how the brain works kind of. if there is a z instead of... its broken keyboard.

[@sajidahmed5079]

keepit up...use beagle bone blue . if using Ai than use Beaglebone AI..it has cloud9 IDE and AWS cloud access . AWS has trained algorithm and Deep learning engine

[@Speedy0475]

Dobra robota,no wiesz robot-a

[@VeerDaVlog]

7:11 - How did you shoot timelapse like this?

[@aymansharaf7456]

Can you give us the modified code!!

[@tharakadimantha1043]

good job brother.Thanak you

[@SadiqueKhan449]

Ultrasonic is not good for corner objects because sound doesn't reflect back

[@kreisumfang6175]

Agradable, me gustaría ver proyectos similares con Arduino y explicados a nivel de código, gracias.

[@ikomtek]

I want to ask, why chinese toy can do this thing without usng AI neural network?

[@bzqp2]

0:57 Nope. They are not "connected together in layers". In classic networks the neurons within a layer are not connected with eachother.

[@dineshvyas]

STM32 is better than Arduino for this kind of processing.

[@joetke]

Arduino features are so minimalist I agree. What could STM32 bring significantly compared to arduino? DMA? on-board RTC? Would appreciate some explanation instead of free statements...

[@kunalsalvi8382]

@@joetke more DMIPS. Faster clock, like you can literally overclock it. And you can code all the math in assembly if you have time and then optimize it equation by equation. I have a 2 years of experience in stm32 and can tell you that once you go ARM you never go back.

[@nevillelusimba1689]

@@kunalsalvi8382 do you mind if i got your contact info? would like to talk about the STM32 and other ARM controllers. I'd like to experiment more with these

[@rooster443]

@@joetke 32 bits against a 16 bit processor, also can overclock his processor up to 4 times than arduino does. Its in the datasheet

[@supermodernotv]

First of all Arduino doesn't mean 8bit processors. We make a lot of boards including CortexM4F processors so. Having said this for the project this guy is doing the UNO is enough. For more complex stuff check out TensorFlow Micro on the SAMD21 boards

[@sajidahmed5079]

use interrupts...based programming and control

[@shivambansal3560]

Well done buddy..!!!

[@MrSaddamko]

As for me - using neural network on Arduino - this is bad solution. Like as run Doom on scientific calculator. Big work with small result. May be using fuzzy logic is better way, it's simplest and more prognosable. Simple rules in rule block, something like "Distance is big nd obstacle is right then rotary_angle is zero", "Distance is low and obstacle is right then rotary_angle is positive" Neural network may be better for big data analyse.

[@joetke]

Totally agree. The title is kind of weird...

[@supermodernotv]

The kid is trying to learn, can't we all just encourage him? :)

[@thorzhou4468]

Great job! Is this project open source, where can I see the source code?

[@abc31905]

To normalne że rozumiem Twój angielski lepiej niż gdybym słuchał anglika?

[@janczaja3965]

Mam podobnie :)

[@nikodembartnik]

Chyba nie masz się czym martwić, ale lekarzem nie jestem :)

[@pedropabloantunezgomez7408]

Could you please share the code modified? I am very interested in working with it with my students but cannot finish the project.Thanks and good job

[@herok4306]

Bro what's the background music...?

[@Tamingshih]

❤great job

[@nikodembartnik]

Thanks!

[@philippebenaim2676]

Nice, but I can see that you don't wait enough time to take the distance with your ultra sonic sensor. So I think you have many errors of measurement... Try Vl53l0x, it's little laser in I2C, and is very fast and accurate 👍And you have many libraries for arduino 😋

[@xianox8]

Is it even viable to train a NN in this way (without it taking for ages)? I'm convinced that it is the processing power that is limiting you, but it is that you do it in real time. What I would attempt to do is to make a computer simlation of the car movement and sensors and train that simulated car.

[@vighneshkrishna1716]

Bro , is it possible to get your mail or any other source to contact you. I got a great project in my mind but actually i’m stuck at a point.

[@mohammadwahba3077]

good job and idea

[@arnabmandal1332]

Good work buddy..

[@nikodembartnik]

Thanks!

[@1400d4d]

Dlaczego ten film nie jest po Polsku ?

[@nikodembartnik]

Ponieważ jest po angielsku

[@Kanjox]

awesome

[@noa7784]

Hello, I really needs some help in path planning. I do hope you can help me. I have the same prototype as you had. Only is that I have to implement path planning with obstacle avoidance.

[@sarojverma6299]

Make sensors in 8 sides and increase their range and it will be perfect also slow the speed

[@farhanamin4075]

Can anybody help me understand the concept of neural network being applied on Obstacle Avoidance robot

[@crayoneater68.5]

That's cool but didnt roomba make this already? What I'm wondering is what benefit the neural network will have for it to determine to go forward or not

[@jay-rathod-01]

Beautiful

[@eUltrabyte]

Dobry film ziomek

[@HeadsUpTV]

Bill Gates, is that you?

[@johnschmidt874]

Yes. He shrank

[@swathikaravindran1942]

How the neural network algorithm to be inserted in to the arduino board

[@hirunawadasinghe]

So is that your Neural Network Robot?