NVIDIA Jetson Nano 2GB is the Best Way to Learn AI
ฝัง
- เผยแพร่เมื่อ 17 ต.ค. 2024
- NVIDIA announced the Jetson Nano 2GB. Affordable (only 59$) version of their edge computer for the AI applications. Is it any good? In my opinion it is really great product if you want to learn AI (as myself). And even it will be perfect for some DIY projects. The only real drawback of this computer is single USB 3.1 port.
Pre-order Now: nvda.ws/30v5w3M
Special thanks to my special Patrons: Wesley Corp, RobotDigg Shanghai, Pierre Freyssinet, M. Aali and Shounak Bhattacharya!
Please subscribe. This will help me to develop other projects like this, to bring the bright future closer!
One time donation:
www.paypal.me/...
If you want to help this channel, please support me on Patreon:
/ skyentific
Instagram: / skyentificinsta
Facebook: / skyentificface
Twitter: / skyentifictweet
#JetsonNano #AI #GTC20
![⭐ฮิตในTikTok!! ( มะมะมะหมูเด้ง MooDeng ) Ver. แดนซ์ ReMix BY [ ดีเจกิต รีมิกซ์ ]](http://i.ytimg.com/vi/4noyAoMa7bQ/mqdefault.jpg)
Your robotic arm was my first electronics/robotics project as a someone new to the field and here I am completing my Masters in Robotics. Thanks!!
Which university brother?
Is it available in India, how can i get harsh bro
Love your videos man. I got a job as a Robotics Engineer and I'm following your content more now than ever.
It blows my mind that you can buy a full computer, designed for machine learning, for $60. We live in the future
Completely agree!!! Now we also need to add affordable robotics to this.
Awesome boards, however the soc of the jetson nano is from the past (based on the Tegra X1, whitch got released in 2015)
@@mampfi isn't everything from the past? unless you have a time machine you're not sharing :/
You can do more with a raspberry pi, th-cam.com/video/aimSGOAUI8Y/w-d-xo.html (and this is a year old!)
@@jameshughes3014 Damn you got me, in my previous comment about the "Jetson Nano Next" (released in 2021) and the "Jetson Orin" (released in 2022) the story of the application engineer is made up to hide my little secret maschine!
Thankyou for taking the time to do this video 🙏🏻 great presentation
love the scenic Sky! fingers crossed 4 a ''new'' nVidia product line!!👍🍌😁
I can't believe that I've been watching almost all of your videos and have never liked nor subscribed. Well, I just fixed that today. Thanks for your great videos.
I frickin love your music choice at 4:10!
Nice! I just bought one. P.S: I recognize that lake and mountains. Living around the Lac Leman too!
:) It is a nice place! I am in Lausanne.
Thank you: I would like to see you build any type of Robotics project with the Nano 2GB.
Thanks a lot , we bought the nano 2
and we are definitely learning AI with it
When you say the model cannot be too complex for this device , can you give a few examples of what a complex vs simple model (roughly) would be like? Thank you all your videos are great!
Как приятно слушать Русскую речь)
You really enjoy what you do and share your enthusiasm, great informative and fun video
PLEASE show how to use this board with a ROS project, this would be AMAZING!
Like any other computer with Ubuntu :)
That mountain view is superb
Many thanks for all you are doing 🙏
Thanks...I really want to see your next project on nano 2GB nvindia kit
*Runs to check oven* Nope, only bread :/
Lucky. Couldn't even find bread in mine. Just pans.
Another great video! Looking forward to hearing more about the board you are developing!
Thank you!
Got plans myself of sticking one in the work truck to be used as an ANPR, seems like a fun project.
Thank God I ordered one already but will need to wait for 4 weeks for delivery.
Please do a project with the Jetson Nano. I bought a Jetson and am looking for fun stuff to try out. Love your channel.
Yes, do some projects with machine learning on the jetson!
Amazing video! And amazing sunglasses!
As ive mentioned on other channels there are a number of solutions for ML/AI work out there, SiPeed M1 and Maixduino ( on the edge ) embedded systems work really well for camera recognition using TF / Keras and Yolo ( v1 and v2 ) also intel's Movidius accelerator for ML/AI is available. Fortunately NV are not the only game in town and most definitely not the most cost effective.
AI/ML is better learned by understanding as in the Math behind it and the language you choose to write your models etc etc, obviously with Python knowing fundamentals like numpy and pandas is essential. Books I would recommend would be the AI/ML Books from O'Reilly there are two here which are very good for the beginner ( Hands On Machine Learning & Tiny ML ) also the 100 paged Machine Learning Book.
Understanding the why will speed up the how thus improve your modelling skills as you go.
what is the most complex thing this can do?? (autonomous driving for a remote car? obstacle avoiding?? or even a walking robot dog?)thx
in other words: what can this be used for?
sorry if my question sounds stupid.
I am a total noob. just found this channel recently and started watching all the robotics stuff
(without any knowledge. starting from scratch)
thx
Your my hero ! Awesome video !
I really love your style !
please make a video about integrating this board with robots, I need this for my project, tnQ in advance, stay safe and may the robot torque be with you :D
Awesome! Can you make more videos on controlling your robot arm using ROS or ROS2? Thanks and please keep up the great work!
Very nice video right here. Done subscribe champion!
Great video as always.
pre-ordered from Amazon. Now I wait :)
Do you already have a video or more information about the board for the Jetson Nano, Xavier and Raspberry Pi you talked about in the video?
I agree. You have to struggle with the limited memory and messy Python packages, you will learn a lot when make some basic things working.
Yes please more videos on this with machine learning.
Спасибо за видео! Отличная работа.
You have all the interesting toys!!! I want to play!!!
I would love to see a project which incorporated the object recognition / localization using the Jetson and your small robot arm. Idea would be to have the robot arm separate parts (bolts, screws...) based on size into containers.
So the board does not have built in wifi?
Nice video, so can I use this in my bioloid premium robot?
Looks like a lot of fun to experiment with.
I’ve bought the original jetson nano developper kit. Is the new one way better and should i refund my purchase ?
Great Video, thanks!!
Super interested in the HAT you showed off! Any more details you can tease about what it's designed to do?
So I know you like the ODrive a lot, as do I, however I've been searching on the community section in the ODrive website and I'm not able to find anything regarding a G-Code Interpreter other than a post saying that they would like to implement that at some future point. Are you aware of any G-Code interpreters available for ODrive?
I've tried to install Tensorflow & Numpy on a Jetson Nano - should be a simple procedure, but ended up going down a rabbit hole. Many others have experienced the same issue
Excellent video!
Is it possible some how that macbook Air M1 can be use an monitor for Jetson board ? ?
Please reply I am new to this field
As far as I know, you cannot use it as monitor, but you can connect to jetson, via special usb port, and you can run programs like this.
@@Skyentific
Thank you so much for suggestions and for prompt reply.
Great tutorial, what's the type of screen you use ?
I want to reject defective product off the line before it jams the next machine in the process. Is this thing powerful enough to analyze 75 widgets per minute?
Looks like the Swiss Alps in the background. Are you in Italy or near Geneva?
Exactly. I am from Lausanne.
@@Skyentific
I had the same feeling of
knowing the landscape and i was right.
So hi from the other side of the lake
Hi @Skyentific, I have an off-topic question, could you give me a suggestion on how I can control up to 16 stepper motors simultaneously by arduino uno with TB6600 drivers, but at different speeds and directions? I thought about using the 74hc595 shift register but I'm not sure which is the better approach. Thanks for your time, my best regards
Good 👍 .. it comes with wifi built in ?
Yes please I vote more learning about NVIDIA Jetson Nano 2GB.
Thank you
Most important to know that you will get a old linux kernel 4.9.140 with no way to upgrade. They not updating it for a few years already
I wondering if this board has enough power to create a quality motion capture system for animation purposes, I we hook up a few depth cameras to it? So at the output we have a stream of data which contains a set of yaw-pitch-roll angles for a set of bones.
Thank you!
I think a traffic control program for street lights would be a pretty nice project :) but then you would need to evaluate four camera streams, maybe by ipcams?
安く容量などが改善されたのは嬉しいですね😊😎😎
can I power jetson nano using 18w power bank?
Spec: It supports 5.1V/2.4A, 9V/ 2A MAX and 12V/1.5A MAX
Normally it needs 5V 3A. But most probably 5.1V, 2.4A would work. In worst case, you can limit the power consumption in Ubuntu.
@@Skyentific thank you so much sir
The Datasheet says 5.25V Max. The Brown-Out Detection kicks in at 4.75V, so the voltage at the module must be whithin 4.75V - 5.25V.
Thank you for this video! Именно то, что мне было нужно
Спасибо за обзор, полезно. Подскажите, пожалуйста, какие подходящие к Jetson Nano 2GB проверенные варианты видеокамер можно купить, чтобы сделать хорошую систему по работе с видео и изображениями?
Вопрос хороший. Я знаю что камеры от распберри пи подходят (начиная от второй версиии) и последняя их версия вроде бы очень даже качественная, можно даже хорошие объективы ставить.
У WaveShare есть для него хорошие камеры.
@@ighor спасибо - решил попробовать waveshare IMX219-160IR
I want to build either jarvas or C-3P0
Nvidia ARM software support is awful. All their devices running Linux kernel version 4.9.140, that is few years old one.
There is no way to upgrade a kernel to a newer version. If you just agree with Ubuntu to do any upgrades, it is break all Nvidia drivers.
And there is no way to install a Wi-Fi 6 drivers (which is builtin in the kernel v5) and not to break all other drivers.
For all old devices they just stopping to release updates, so you will stuck on some kernel version forever.
you should also try google Coral, it seems as powerful as xavier and you can use rpi4 as host.
Sir i need your help I have no Idea of programming and I want to leran ai and this decision making field can u help me which degree or programming language to learn
Hi, I have subscribed like long ago and appreciating your videos. I am wondering if this Jetson Nano can go wireless. Like can it be on separate robot and communicate to the computer wirelessly?
Thank you!
Soowhan Yi yes, of course. You can communicate via ssh over WiFi. I tried. It works. :)
Thank you for subscribing, and for watching!
@@Skyentific Thank you for the reply. Now I have follow up question. If I needed to use two real sense camera, does that mean that I have to use two Jetson Nanos? Is there a way to use those two cameras at the same time? Again Thank you so much for your reply!
Nvm. I just need to get the upgraded version. lol
Learn Ros
The question is can you train AI using those boards or you have to pay a lot of money on the cloud to be able to use those boards
Some of the nvidia courses are free. It actually only looks like one of them.
I have a GTX 1070 card, do I still need this? Thanks.
So this is Justin Nana. Interesting
Жаль что nvidia Jetson nano нет в наличии на официальном сайте nvidia. Интересно, если заказать nvidia Jetson nano с aliexpress, то пришлют качественный товар?
Your business is pioneering to wish to work with you
Great job as always!! You can check out robot ignite academy to learn ros its not free but they teach it pretttty good ive been learning it and they made it seem pretty easy for mee. Good content as always.
How good is this one compared to the 4 gb memory-size ? I mentioned this since most tutorials on this subject are carried out with the largest memory size Jetson nano. I will have to work on a collision avoidance project for school but haven't decided yet which one to acquire.
Not easy to answer. The 4GB version uses about 1.6GB of 4GB during idling so I guess that the 2GB version uses 75% of the memory even if it's just idling, leaving 400Mb free, whitch might not be enougth for some applications. Maybe the nvidia guys implemented some tricks to reduce memory consumption, I don't know. If you plan to do any kind of neuronal network training on the jetson, choose the 4GB version. If you run into memory problems there are a few things you could do:
- disable GUI (saves about 600MB)
- increase the swap
Can you suggest some books on electronics for robotics?
You can use LINO robot in github.
I am waiting you to build mobile robot based on ROS navigation and including COBOT for pick and place task.
For ROS its easy you need Lidar and go to make mkdir CAtkinws catkinmake bash/devel and other codes
Can you add a CAN bus shield to the Jetson Nano ?
Seems like this is not possible to preorder in Canada, as usual for nVidia
You can by it now on Arrow electronics website. It’s in stock there.
@@diggleboy I actually ended up doing exactly that. Though it will be annoying having to pay the brokerage fee.
I'm trying to calculate a price ahead of time just in case. $59 US and about a 33% exchange rate... ya, shoud be about $200 CAD.
Please use a black PCB
pls make a deeplearning project video. I was thinking about a döner filling robot arm, maybe you could do that :P
what do you do for work?
what about books that we can read to learn robotics/ai?
Craig, John J. "Introduction to robotics: mechanics and control", 3/E. Pearson Education India, 2009.
Jazar, Reza N. "Theory of applied robotics: kinematics, dynamics, and control." Springer Science & Business Media, 2010.
@@YetiFell ty for the reply!
Is the 2gb Jetson good enough for learning AI? The reason that I am asking is because the price of the 2gb Jetson is the same as the Raspberry pi 4 with 4gb.
Thank you.
It should be. Otherwise, what would be the point of it?
Machine learning video would be good.
The like button is just not enough for this video
Nice...
How is this different from a Raspberry Pi?
It has GPU, which helps with machine learning. For most AI projects it will outperform Raspberry pi.
I love Robotics
i am a developing machine learning algorithms to be used in embedded systems, and i have worked with the jetson nano before. if you want to get into deep learning, i'd rather recommend buying a gtx 1070 graphics card and work on a PC. You will not do training on the Jetson Nano. You don't want to, i promise you that. While the Jetson Nano is indeed powerful and can run ML algorithms, the fact that it's a arm64 processor means that you won't have an as broad compatibility spectrum with different versions of ML libraries, which you might need for interesting projects you found on github. It's an extra hassle going through the processes to be able to run them on an arm64 based machine. if you want to get into machine learning, you want to train your own neural networks, and a gtx 1070 equipped PC is far better suited for that than a system that puts certain restraints on what it can do just so it has a small form factor and is a single board computer.
One is $60 and one is $339, what did you expect?
Спасибо, за видео!!!
Спасибо за комент, и спасибо что смотрите!
@@Skyentific Очень интересные Ваши видео, помог обзор NVIDIA теперь более ясно что умные устройства можно делать не только на Arduino and Raspberry, и наконец то !наглядный! пример как работает камера! понимается принцип ROS
The price is not the limiting factor, grey matter is the problem. It's all so so stupidly hard, the maths, inverse kinematics the Jacobian matrix, epipolar constraints, homogeneous cords, gaussian image smoothing, SURFS and so on.
In the UK people who try to specialize in in fields like this and or astronautics engineering are called semiskilled or unskilled people, because engineering is an unskilled at best vocational career.
@Skyentific
I realize that this is a very cheap board to work with and is great for entry level! But in the past when using the Nvidia Jetson v1, we had a lot of driver compatibility issues with various programs and hardware due to the ARM architecture. Is this no longer a problem? What about the UDOO x86 computers? (shop.udoo.org/products/x86.html) What are your thoughts on this computer as an entry level robotics computer?
the main advantage of NV boards is GPU with a lot of PU. And now NVIDIA is introducing DPU just for data...
38 seconds into the video and I can't stop wondering:
Или автор русский и ничегошеньки не хочет делать с акцентом or if this is the most hardcore russian accent imitation I've ever seen.
Good vid though
Или автор так давно говорит с этим акцентом, что ему уже больше ничего не поможет!
@@Skyentific Ролик досмотрел, несмотря на то, что к роботике или МЛ не имею никакого отношения. Отношение, подача и энтузиазм очень понравились. Кип ап зе гуд ворк, комрад. Айм рашен, ай донт лайк энибади. Бат ай дринк ту ю! (С) Ред Алерт
please make some ML vids with the jetson!
Couldn’t you just use an usb-hub? :)
Engineering Spare Time I am not sure, as RealSense camera sends a lot of data. I am not sure, they will fit in one usb, even with the hub.
Skyentific It can handle 5 Gbit/s.. There is ‘just” one way to find out :)
108€ Jetson Nano kit is now 189€. Screw this timeline.
Спасибо. Уже заказал.