Jetson Nano Developer Kit - Getting Started with the NVIDIA Jetson Nano

today in the workshop we'll be looking at the Nvidia Jetson Nano a development platform for artificial intelligence after examining the nano platform we'll install Ubuntu Linux 1804 and the Nvidia Jetson SDK it's an intelligent episode today so welcome to the workshop [Music] hello and welcome to the workshop today we're going to be looking at an artificial intelligence experimenters platform now this is the device over here this is the Nvidia Jetson Nano and it's part of the Nvidia Jetson series of artificial intelligence development boards now you may not be familiar with the other Jetson series boards and that's not too surprising they're rather expensive they range from about three hundred dollars to about 1,100 bucks but this one came does everything this is a $99 board that brings all the power of AI down to a level or experimenters like us and students can afford it now there is a lot that you can do with this for this board is based on the same technology as the other Jetsons and it can use a lot of the same software specifically the software included in the jetpack utilities provided by Nvidia now what we're going to do today is we're going to take a look at this board we're going to basically install an operating system on it get familiar with its features and then we're going to be using it in other experimenter projects now as you can see this board looks a lot like a single board computer but as you're about to discover it's a lot more than a single board computer and as you'll also notice I seem to have two of them the reason for that is I'm going to be incorporating this onto my db1 robot as well but before we can do anything with the Nvidia jets and we need to know a little bit about it so let's go and take a look at the Jetson right now and see exactly what this platform is the Nvidia Jetson nano developer kit was announced in late 2018 and was released in the first quarter of 2019 the kit retails for 99 US dollars and is available from Nvidia as well as a number of distributors including Pine Maroney Amazon and Sparkfun the kit consists of the Jetson nano module and a carrier board it uses a micro SD card with the NVIDIA jetpack SDK software the Nvidia Jetson nano is part of invidious Jetson family of AI development boards now here are some of the Jets and development platforms there is the Nano of course which we're looking at today it has a power of 0.5 teraflops consumes 5 to 10 watts and is priced at $99 next up is the TX 2 it has a power of 1.3 teraflops consumes 7 to 15 watts and is priced from $2.99 to 749 u.s. dollars and then there is the gets an ATX Xavier it has a massive power of 10 teraflops consumes 10 to 30 watts and is priced at 1099 dollars the deaths in nano is capable of running multiple neural networks in parallel it's useful for image classification and segmentation it can also be used for object detection and can be used for speech processing the Nano is based upon the Nano module which is available separately the module uses a 260 pin s Odom connector the production versions of the Nano module use nvram instead of a micro SD card the specs of the Nano are impressive it's a quad-core arm a57 1.4 3 gigahertz CPU it has a 128 core nvidia maxwell vpu there are 4 gigabytes of high-speed memory it has a 40 pin GPIO bus plus I to see SPI and UART connections there are 4 USB 3 ports and a Gigabit Ethernet port it has a CSI video camera interface it's compatible with the Raspberry Pi version 2 camera it has HDMI and DisplayPort outputs each support 4k video and they can both be used simultaneously there is an m dot 2 key ii socket for a Wi-Fi and bluetooth adapter you supply the Jetson Nano with 5 volt power using a micro USB or 2.1 millimeter jack so now let's start working with the Jetson Nano so here's the Jetson Nano development board now I've got it all hooked up but I'm going to unhook it so we can take a better look at it so I'm just going to disconnect this which is the power supply an Ethernet cable a USB which in my case is going to a keyboard and a mouse and an HDMI and we'll take the Nano board off now before we look at the board I just want to show you the box the box came with this little stand which is rather amusing you place that in here and then you can just place the board on here and it gives you a convenient place to work on the board which a chess is pretty cool now let's just take a look at this side of the board over here as you can see we have a series of jacks on there the ethernet that's a Gigabit Ethernet there are four USB three jacks there is a HDMI and a DisplayPort output and these can both be used Dainius Lee by the way both at 4k video and over here is a barrel connector that you can use to supply 5 volts at up to four amperes if you want a higher ampere supply now this connector over here the microUSB is only used for it up to two and a half ampere supply so if you want to use this in high powered mode you're going to need to use this power supply and when you do use that power supply there is a jumper here that you're going to have the jumper in order to take power from the barrel connector instead of the microUSB connector now down on here is a CSI connector not certain how well you can see that that's the connector for the Raspberry Pi camera there's a few other i/o connectors over here power over ethernet connector down here down this side you'll see a 40 pin connector now that's a 40 pin GPIO and its output pins are programmable but by default this is the same GPIO that the Raspberry Pi uses so you've got the ability to use some Raspberry Pi designs with the Nano although of course you won't be able to put a hat on top of here because of the heatsink now the heat sink itself sits on the nano module the Nano module itself is an Esso dim module it's mounted on a socket I don't know if you can see the socket over here and when you remove it and you'll have to take these two screws off to remove it there is another connector underneath that's an m2 type connector that can be used to add a Wi-Fi and Bluetooth connection to the Nano now you'll also notice that the heatsink has four threaded holes in it this is for mounting a 40 millimeter fan if you want and this connector over here is for a 5 volt 40 millimeter PWM controlled fan so that's where you would connect it and so that is the NVIDIA Nano in a nutshell the Jetson Nano so now that we've seen what the Nano is I want to discuss for a few moments what the Nano is not now if you take a look at the Nano that I've got here on my workbench you can see that it is a board that I have an HDMI cable I've got a USB cable that's going to a keyboard and a mouse but an Ethernet cable as well as a power supply connection and by looking at that you would be forgiven for saying well this is a single board computer well it's not a single board computer for one thing it's not a single board it's a module sitting on top of the carrier board and while that may seem to be a small distinction it's actually a very big one the idea with the Nano is that you can buy the Nano modules separately and the module you buy separately does not have a microSD card on it instead it has NVRAM on it the idea is that you use this platform to develop code for the Nano module using the microSD card and then you burn that code onto the NVRAM because the Nano is a developer's platform it is not a single board computer I seen a lot of people criticizing it because it doesn't have Wi-Fi and Bluetooth I can understand the criticism after all a $10 Raspberry Pi 0w man kids to squeeze Wi-Fi and bluetooth onto it for that price why can't the Nano at ten times a price have it as well well again the Wi-Fi and Bluetooth is not on the module it's on the carrier board and the carrier board does a place for a separate m2 module that you can add Wi-Fi and Bluetooth with but it is not a single board computer it is a development platform and if you use Wi-Fi and Bluetooth in your application then you would add that module to your application same deal with the video ports it's got both DisplayPort and HDMI but if your application doesn't need a video output you wouldn't build that into your product but you would put an Esso dim socket on your product to accept the Nano module so now that we've got that out of the way let's get some software installed onto our Nvidia Nano now invidious provided an excellent resource for getting started with the Nano and I've got a link to that resource in the article that accompanies this video the resource goes over the different components of the Nano and goes through what you're going to need to prepare for setup so in addition to the Jetson Nano itself you're going to need a micro SD card now they recommend a minimum of 16 gigabytes but I would say that you want to use at least 32 gigabytes if not more you'll also need a micro USB power supply that can supply at least two and a half amperes not all of them can but you'll need that the power of the board of course you're also going to need a keyboard a mouse and a video monitor as well now in the next step you write the image to the micro SD card and this is where you get the image from over here the Jetson nano Developer Kit SD card image now you're going to want to download that now this is a large download it's over 6 gigabytes so if you're on a slow internet connection this might take a bit of time and videos also supplied instructions for Windows for Mac and for Linux for burning the card although the basic instruction is as follows you get the image and then go and use an SD card burner now if you don't have an SD card burner the one that I would recommend is XAR and you'll also find link to that in the article accompanying the video I've used extra before in several videos to learn micro SD and SD cards so I won't show you it but it's very simple you to select where the images select the card you want to burn it to and then let it go and do its business and etcher is available for linux for mac and for windows so once you get the image onto the card it's time to place the card in to the Nvidia detsen and we can work on the installation now so I've got my nano hooked up again I'm using the micro USB power supply in this particular case it should be ok to bring the board up eventually I'm going to move to that power supply however I've connected an Ethernet cable this is important to do when you're starting everything up and you're going to also need to connect the keyboard and a mouse and you can use the four USB jacks over here for it I only have one connection because my keyboard and mouse are already connected together through an active USB hub on my workbench and I've also connected a video monitor now I've used an HDMI monitor you could also use a DisplayPort monitor and is connected into the appropriate connector and then naturally you're going to need to put in your microSD card now the way the microSD card goes is you take the contacts and have them up toward you and if you look at the bottom of the nano module and am not sure if you can see that too easily you will find the actual microSD slot and you'll just put the card into there and again I hope you can see that a black card on a black background and you press it in and then the card will be installed and so once you've got that micro sd card in it is time to get our nano working now during the installation of the software you'll see a lot of files being extracted and installed it's essentially identical to the installation of Ubuntu Linux because that is what we are installing this just happens to be a customized version eventually lend up on the system configuration screen over here where they ask you to accept a term of the License Agreement so we will do that and continue we'll select our language and we'll select a keyboard type it asks where we are so that it can establish what timezone were in Toronto is close enough to where I am so I'm going to continue and now we have to give it some identification give it the computer a name and pick a user name and Cleaves a password and you can choose whether to log in automatically or require a password to log in look a lot of people may find it convenient to log in automatically I'm from the old school which says I always need two passwords all set it this way and now it is going to start to install a number of other files as well and complete the installation and I'll skip the rest of the file install and just move ahead to the final results after the files install and a number of reboots you will finally end up at this screen which is a standard Ubuntu 18:04 desktop Nvidia has added a few things to the desktop they've added the chromium web browser they've also added a couple links to the developer zone and to the support files online otherwise this is essentially the same desktop that you would get with any installation of Ubuntu 1804 so if you're familiar with the Ubuntu desktop you're going to be familiar with this environment and so that essentially is the installation of the software for the Nvidia Jetson Nano so now that we've set up our Nano what exactly can we do with it well NVIDIA has a number of resources up on their website including hello ai world which is a series of steps that you can go through in order to demonstrate some of the capabilities of the Nano they have demonstrations for things like image recognition and object classification and hello ai world is a great suite of tools to go through and the next time that we get together and talk about the Nano I intend to do exactly that and go through them so I can show you the AI capabilities of this device but right now we don't really have the time to do that hello ai world is going to have to be the subject of an entire video because there is so much to it however I thought it would be a good idea to test the capabilities of the Nano so I wanted to run at least a couple of demos in order to do that one other thing I wanted to mention about setting up the Nano is the size of the micro SD card I neglected to mention that before now if you look in the documentation for the Nano it will specify a minimum of a 16 gigabyte micro SD well 16 gigabyte is really not a realistic assumption you need to use at least 32 in fact what I've used over here is both of these cards and they both worked very well for me I've got a 64 and a 128 gigabyte card neither both kinks and cards I heard a few people on the forums saying specific SD cards weren't working well for them these kingston cards seem to work pretty good for me i'm not advertising kingston but they were a good value the cost difference between the 64 and the 128 gig card was only about five US dollars so if I were you I'd go for the 128 gig and leave yourself a lot of room for experimenting so let's take a look at a few things now that we can do with our Nvidia detsen nano so let's take a look at what we can do to demonstrate the power of the Nvidia Jetson Nano now included with the installation are a number of different applications and example programs now these actually need to be built you can't just go and run them and I've built a number of the CUDA samples now cuda is a parallel processing platform that nvidia has created and they have a number of sample applications as i said you have to build these and the build process takes a little while it takes about two and a half hours or so to run it on the Jetson Nano and one thing I'd advise when you do this is keep your hand off the heatsink because it actually gets very very warm once you've completed the build process and I have instructions for doing the build process and the article accompanying this video if you want to do this as well you will find that there are a number of different demonstrations over here now some of these demonstrations simply do mathematical calculations and come up with numbers and they can be used for benchmarking other ones are more graphical in nature and as you can see there are quite a few of them I was playing around with them and I found three of them that I thought were amusing so let's give them a try here's one called ocean fft fast fourier transform and here's our ocean and we can use our mouse to to make our ocean move around let's make it a bit bigger and I can move the application around and I can zoom in zoom out etc now remember every one of these points is being calculated right now so this is basically a test of raw computing power ok let's try another demonstration here smoke particles now this is an interest in one and in this demonstration as the particles dispersed you can use your mouse to move around the room and you'll see the shadows of the particles as reference to the light source which is behind the particle barrier to see them the light source right now but again every dot over here is being calculated and so it's a illustration of how much data this can calculate in a very short period of time okay and we've done smoke particles oats and loads the random fog there we go now in this one you use the keyboard to manipulate it and they give you instructions over here for doing it and so you can if you want to you can zoom this ten times again each one of these points is actually being calculated the lines represent the axis that this is going in and you can use some of the keys on your keyboard change the rotation I'm changing the access over here again this is a representation of how many calculations this can make in a given period of time okay so those are this three of the examples that I wanted to show you there are as you saw many many more of them you can also try doing things such as running video etc onto the platform I I've tried a few YouTube videos everything is very fast and responsive and as a single-board computer even though it isn't one it would actually be a very good machine for $100 but of course we know it's capable of much much more so that wraps up our look at the Nano for today but by no means is it's the last time that we are going to be seeing the Nvidia Jetson nano now I have as I said bought two nanos and one of these nanos is intended for my build a real robot project so for those of you who are following the construction of the DB one robot I will be using an Nvidia Jetson Nano as part of that design so you'll be seeing a lot of that later now for those of you dis following the workshop videos the next time that we get together to talk about the Nano I'm going to be running the hello AI world applications that NVIDIA has provided now these are a set of demonstrations that will show you some of the capabilities of the Nano some of its object detection and object recognition capabilities are shown in hello AI world and we're going to see how we can train our nano to recognize specific objects now after that we've also got to install some extra hardware onto the Nano so I'll be doing another video showing you how to set up both the Bluetooth and a Wi-Fi module as well as the PWM controlled fan that can bolt on to the heatsink and finally there is a really cool Nano project that I want to make use of it is something called a jet bot now the jet bot is a small robot cars a small two-wheel robot car it's completely open source you can build it with essentially a Jetson Nano a little power supply for it a little USB supply a Raspberry Pi 2 video camera and a robot car base so I'm going to be putting one of those together now as I said all of the code for the jet bot is open source so you can look it up and start building one yourself right now they also give us TL files if you want to print a 3d chassis for it but I'm going to just use a standard robot car chassis so there is a lot that we are going to be doing with the Nvidia Jetson Nano in the future now if you want to stay informed about all those videos the best way to do that is to subscribe to the YouTube channel so please hit the subscribe button I would very much appreciate that and as always you will find an article accompanying this video on the drone bot workshop comm website there's a link below the video to that and when you're on the website please sign up for the newsletter so we can stay in touch with each other so until we meet the next time please take care of yourselves and I hope to see you again very soon here in the workshop good bye for now [Music] you [Music]

Channel: DroneBot Workshop

Views: 118,148

Keywords: jetson nano, nvidia jetson nano, nvidia jetson nano developer kit, ai development, artificial intelligence

Id: BkZ1n_1F-Cg

Length: 24min 57sec (1497 seconds)

Published: Sat Jun 01 2019