# Jetson Published articles for Jetson. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## TensorRT Edge-LLM Completes the MLPerf Edge Agentic Benchmark 6.4x Faster on Jetson AGX Thor DevFeed: [TensorRT Edge-LLM Completes the MLPerf Edge Agentic Benchmark 6.4x Faster on Jetson AGX Thor]() Original publisher: [Read original article]() Author: Elizabeth Goodman Published: 2026-09-16T20:37:07Z Content type: article Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [Jetson AGX Thor Developer Kit](), [Inference Performance](), [benchmarking](), [TensorRT](), [Large Language Model](), [quantization](), [Artificial Intelligence]() Tags: [agentic-ai-generative-ai](), [ai-agent](), [ai-inference](), [evaluation](), [jetson](), [jetson-agx-thor-developer-kit](), [llm](), [llm-benchmarking](), [mlperf](), [nvidia](), [performance](), [quantization](), [tensorrt](), [tensorrt-llm](), [thor]() ### AI overview This article reports that NVIDIA TensorRT Edge-LLM ran Qwen3.6-27B on a single NVIDIA Jetson AGX Thor Developer Kit for the MLPerf Inference v6.1 Edge Agentic benchmark. Using NVFP4 quantization, tree-based multi-token prediction, and KV cache reuse, it achieved 52.33 tokens per second and completed 1,007 turns in 24 minutes and 36 seconds, 6.4 times faster than the llama.cpp reference submission. ### Source excerpt AI agents are moving from cloud data centers to vehicles, robots, and other edge devices. Unlike a chatbot that answers a single prompt, an agent works through... ## Frontier Reasoning Reaches the Edge: How to Deploy and Optimize Models on NVIDIA Jetson DevFeed: [Frontier Reasoning Reaches the Edge: How to Deploy and Optimize Models on NVIDIA Jetson]() Original publisher: [Read original article]() Author: Elizabeth Goodman Published: 2026-09-04T16:21:04Z Content type: tutorial Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [Jetson](), [model-deployment](), [Inference](), [Inference Performance](), [Optimization](), [Mixture of Experts (MoE)]() Tags: [agentic-ai-generative-ai](), [edge](), [edge-computing](), [how-to](), [inference](), [inference-performance](), [jetpack](), [jetson](), [jetson-orin](), [mixture-of-experts-moe](), [nvfp4](), [optimization](), [physical-ai](), [reasoning](), [thor](), [tutorial](), [vllm]() ### AI overview A tutorial on deploying and optimizing compact reasoning and agentic AI models on NVIDIA Jetson. It covers choosing models, improving inference with NVFP4 quantization and speculative decoding, serving example models with vLLM, and validating a configuration for a workload. ### Source excerpt Running reasoning and agentic AI at the edge has been harder than it needs to be. Until recently, models capable of multi-step reasoning were too large to run... ## D-Robotics RDK S100P - A 128 TOPS alternative to NVIDIA Jetson Orin NX 16GB with Cortex-A78AE/R52+ cores DevFeed: [D-Robotics RDK S100P - A 128 TOPS alternative to NVIDIA Jetson Orin NX 16GB with Cortex-A78AE/R52+ cores]() Original publisher: [Read original article]() Author: Debashis Das Published: 2026-08-31T08:06:13Z Content type: news Language: en Sources: [CNX Software - Embedded Systems News]() Topics: [Jetson](), [Robotics](), [Embedded Systems](), [Microcontroller](), [Jetson Orin](), [SOC](), [cpu](), [Arm](), [real-time](), [GPU](), [Linux](), [Nvidia]() Tags: [arm](), [artificial-intelligence-ai](), [camera](), [computer-vision](), [cortex-a78](), [cpu](), [deep-learning](), [development-kit](), [dfrobot](), [edge-ai](), [embedded](), [embedded-systems](), [gpu](), [hardware](), [jetson](), [jetson-orin](), [linux](), [mcu](), [networking](), [nvidia](), [os](), [processors](), [real-time](), [robotics](), [ros](), [safety](), [sensor](), [sensors](), [single-board-computer](), [soc](), [software](), [som](), [specifications](), [systems](), [ubuntu]() ### AI overview The D-Robotics RDK S100P is a robotics single-board computer featuring a 6-core Cortex-A78AE CPU, 4-core Cortex-R52+ MCU domain, 128 TOPS INT8 Nash BPU, Mali-G78AE GPU, and 24 GB of LPDDR5. Its MCU supports real-time motor and sensor control, potentially reducing the need for a separate real-time controller depending on the robot's requirements. ### Source excerpt D-Robotics RDK S100P development kit is a robotics SBC built around an S100P-based module providing an alternative to NVIDIA Jetson Orin NX 16GB with a 6-core Cortex-A78AE application cluster, a 4-core Cortex-R52+ MCU domain, a Nash BPU rated at 128 TOPS INT8, a Mali-G78AE GPU, and 24 GB of LPDDR5. The four R52+ cores can be configured for reliable real-time control: two cores run in lockstep for safety, and the other two support split-lock operation. In practice, the A78AE CPU and BPU can handle Linux and vision processing, while the MCU handles time-critical tasks such as motor and sensor I/O. This can reduce the need for a separate real-time controller, although whether it eliminates one depends on the robot and its requirements. RDK S100P specifications: SoC - D-Robotics S100P CPU - 6x Arm Cortex-A78AE @ 2.0 GHz (safety-capable AE cores) MCU - 4x Arm Cortex-R52+ @ 1.2 GHz (1x DCLS [...] The post D-Robotics RDK S100P - A 128 TOPS alternative to NVIDIA Jetson Orin NX 16GB with Cortex-A78AE/R52+ cores appeared first on CNX Software - Embedded Systems News. ## Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control DevFeed: [Post-Train NVIDIA Cosmos 3 Edge for On-Device Robot Control]() Original publisher: [Read original article]() Author: Michelle Horton Published: 2026-08-19T16:00:00Z Content type: tutorial Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [Cosmos](), [AI, ML & Data Engineering]() Tags: [agentic-ai-generative-ai](), [cosmos](), [deployment](), [edge](), [edge-computing](), [featured](), [foundation-models](), [inference](), [jetson](), [latency](), [nvidia](), [on-device](), [physical-ai](), [post-training](), [robotics](), [robotics-simulation](), [robots](), [simulation-modeling-design](), [thor](), [tutorial]() ### AI overview A tutorial on post-training NVIDIA Cosmos 3 Edge as an on-device robot manipulation policy, serving it on Jetson Thor, running receding-horizon inference, and evaluating it in closed-loop simulation. ### Source excerpt Robots need policies that can adapt to their sensors, environments, and tasks while running on onboard computing hardware. World models offer a foundation for... ## NVIDIA JetPack 7.2.1 Adds Agentic Video Skills and T3000 Emulation DevFeed: [NVIDIA JetPack 7.2.1 Adds Agentic Video Skills and T3000 Emulation]() Original publisher: [Read original article]() Author: Elizabeth Goodman Published: 2026-08-11T19:00:00Z Content type: article Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [Jetson](), [SDKs](), [CUDA](), [GPU](), [AI Inference](), [Inference Performance](), [Latency](), [Low Latency](), [C++](), [Python](), [Artificial Intelligence](), [Automation]() Tags: [agentic-ai-generative-ai](), [ai-inference](), [automation](), [c-plus-plus](), [cuda](), [edge-computing](), [featured](), [gpu](), [inference](), [jetpack](), [jetson](), [latency](), [low-latency](), [nvidia](), [python](), [robotics](), [robotics-compute](), [sdks](), [video-analytics](), [video-codec-sdk]() ### AI overview NVIDIA JetPack 7.2.1 adds PyNvVideoCodec 2.2 support on Jetson Thor, enabling Python-based hardware video encoding and decoding with GPU-resident frames. It also introduces agentic video skills that turn developer goals into device inspection, configuration, execution, measurement, and evidence-driven codec workflows. ### Source excerpt Video is a core data path across NVIDIA Jetson applications, from robotics and intelligent video analytics to industrial automation, healthcare, media... ## Run Local Agentic AI Workflows with Meta's Muse Glimmer on NVIDIA DevFeed: [Run Local Agentic AI Workflows with Meta's Muse Glimmer on NVIDIA]() Original publisher: [Read original article]() Author: Michelle Horton Published: 2026-08-10T13:27:19Z Content type: article Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [Artificial Intelligence](), [Local AI](), [Large Language Model](), [long-context](), [Inference](), [Meta](), [Open Source](), [GPU](), [NVIDIA DGX](), [NVLink](), [Jetson](), [Automation]() Tags: [agentic-ai](), [agentic-ai-generative-ai](), [ai](), [automation](), [context-window](), [data-center-cloud](), [dgx-spark](), [dgx-station](), [edge-computing](), [featured](), [gpu](), [inference](), [jetson](), [local-ai](), [nemoclaw](), [nvidia](), [nvidia-dgx](), [nvlink](), [open-source](), [top-stories]() ### AI overview Meta's Muse Glimmer is a 30B open-weight dense model designed for local agentic AI workflows. With a 120K+ context window and performance of up to 20K tokens per second on a single GPU, it supports sustained, multi-step tool use and local processing of sensitive data. ### Source excerpt Meta returns to the open source ecosystem with the release of Muse Glimmer, a 30B open-weight dense model with a 120K+ context window built for local AI... ## Build a Multi-Camera 3D Tracking Application with NVIDIA DeepStream 9.1 Skills DevFeed: [Build a Multi-Camera 3D Tracking Application with NVIDIA DeepStream 9.1 Skills]() Original publisher: [Read original article]() Author: Elizabeth Goodman Published: 2026-07-15T23:00:00Z Content type: tutorial Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [multi-camera tracking](), [Nvidia](), [AI Development](), [Jetson](), [GitHub](), [Open Source](), [real-time]() Tags: [3d](), [agentic](), [agentic-ai-generative-ai](), [ai](), [ai-development](), [analytics](), [applications](), [automation](), [computer-vision-video-analytics](), [deepstream](), [developers](), [development](), [edge](), [featured](), [github](), [jetpack](), [jetson](), [metropolis](), [multi-camera-tracking](), [nvidia](), [real-time](), [research](), [spaces](), [video](), [video-analytics]() ### AI overview This article introduces NVIDIA DeepStream 9.1 capabilities for building multi-camera 3D tracking applications. It explains how AutoMagicCalib and Multi-View 3D Tracking combine detections from auto-calibrated cameras in a shared 3D coordinate system, maintaining consistent object IDs across views. It also highlights agentic skills, JetPack support for Jetson edge platforms, and open-source reference implementations. ### Source excerpt Developers building video analytics applications across large spaces must track the same object as it moves between camera views. Single-camera 2D tracking... ## What's new in Swift: March 2026 Edition DevFeed: [What's new in Swift: March 2026 Edition]() Original publisher: [Read original article]() Author: Dave Lester Published: 2026-03-31T14:30:00Z Content type: article Language: en Sources: [Swift.org]() Topics: [Swift](), [cross-platform](), [releases](), [Tooling](), [Linux](), [Windows](), [systems](), [Containers](), [Concurrency](), [Jetson](), [real-time](), [ASP.NET Core]() Tags: [ai](), [announcement](), [concurrency](), [cross-platform](), [developer](), [ecosystem](), [jetson](), [linux](), [news](), [open-source](), [platform](), [programming](), [real-time](), [releases](), [swift](), [systems](), [tooling](), [windows]() ### AI overview A March 2026 digest of Swift project and community news, led by Swift 6.3 and the ongoing integration of Swift Build with Swift Package Manager. It also highlights cross-platform build support, open-source package testing, community talks, systems programming, real-time computer vision on NVIDIA Jetson, a Vapor-based AI data pipeline, Swift Concurrency, and API deprecation practices. ### Source excerpt Welcome to "What's new in Swift," a curated digest of releases, videos, and discussions in the Swift project and community. Swift 6.3 has been released, expanding Swift into new domains and improving developer ergonomics. A highlight of its release is work to improve cross-platform build tooling. Owen Voorhees shares an update on that effort: Hi! I'm Owen, a lead engineer on the Core Build team at Apple. Last year we shared our goal to bring Swift Build to Swift Package Manager, in an effort to deduplicate build technologies within the Swift ecosystem and deliver a consistent build experience across all platforms that Swift supports. Since the announcement, we've been working in the open, landing hundreds of patches to improve Swift Build's support across various platforms including Linux and Windows, and to integrate it deeply in Swift Package Manager. With Swift 6.3, developers have the option to enable this integration and try it out with their packages. To validate parity with the previous build system, we've used the package list from swiftpackageindex.com, testing thousands of open source packages with Swift Build. Most recently, the main branch of Swift started using Swift Build as its default build system, paving the way for Swift Build to be the out-of-the-box option for Swift developers in a future Swift release. Over the coming months, we'll continue sharing our progress and driving down the remaining bugs to bring the build system to parity. We encourage you to give it a try and file bugs that you encounter. We're excited by this progress, and look forward to building future tooling improvements across all platforms and project models that will benefit from this build system. Now on to other news about Swift: Videos to watch Interested in Swift for systems programming? The -ization of Containerization, presented at SCaLE, covers the Containerization project and their experience adopting Swift. Swift community meetup #8 featured two talks: real-time compu ## Building intelligent physical AI: From edge to cloud with Strands Agents, Bedrock AgentCore, Claude 4.5, NVIDIA GR00T, and Hugging Face LeRobot DevFeed: [Building intelligent physical AI: From edge to cloud with Strands Agents, Bedrock AgentCore, Claude 4.5, NVIDIA GR00T, and Hugging Face LeRobot]() Original publisher: [Read original article]() Author: Arron Bailiss Published: 2025-12-12T21:03:04Z Content type: article Language: en Sources: [AWS Open Source Blog]() Topics: [AI Bots]() Tags: [ai-agents](), [amazon-bedrock](), [amazon-bedrock-agentcore](), [artificial-intelligence](), [best-practices](), [cloud](), [edge](), [embedded](), [generative-ai](), [inference](), [jetson](), [latency](), [nvidia](), [open-source](), [physical-ai](), [robotics](), [strands-agents](), [technical-how-to]() ### AI overview An AWS article explores physical AI agents that combine cloud reasoning with low-latency edge execution. It demonstrates a robotic arm and a quadruped robot controlled through Strands Agents on NVIDIA Jetson hardware. ### Source excerpt Agentic AI systems are rapidly expanding beyond the digital world and into the physical, where AI agents perceive, reason, and act in real environments. As AI systems increasingly interact with the physical world through robotics, autonomous vehicles, and smart infrastructure, a fundamental question emerges: how do we build agents that leverage massive cloud compute for [...] ## Turing Pi 2 Supports Raspberry Pi Compute Modules and NVIDIA Jetson Boards DevFeed: [Turing Pi 2 Supports Raspberry Pi Compute Modules and NVIDIA Jetson Boards]() Original publisher: [Read original article]() Author: Techno Tim Published: 2022-01-01T13:00:00Z Content type: article Language: en Sources: [Techno Tim]() Topics: [Hardware](), [Raspberry Pi](), [Jetson](), [Nvidia](), [Artificial Intelligence](), [Machine learning](), [Homelab]() Tags: [ai](), [arm](), [hardware](), [homelab](), [jetson](), [ml](), [noctua](), [nvidia](), [pwm](), [raspberry-pi](), [raspberry-pi-compute-module](), [samsung]() ### AI overview The article introduces the Turing Pi 2, a compact ARM cluster designed for scalable edge computing. It describes its 32GB of RAM, SATA III interface, support for Raspberry Pi Compute Module 4 and NVIDIA Jetson boards, and the ability to combine Raspberry Pi and Jetson systems for general compute, AI, and machine-learning workloads. ### Source excerpt The Turing Pi 2 is a compact ARM cluster that provides a scalable computing on the edge.The Turning Pi 2 comes with many improvements over the Turning Pi 1.This model ships with 32GB of RAM, SATA III interface, Raspberry Pi Compute module 4 support, and support for NVIDIA Jetson boards.This means that you can mix and match both raspberry Pis along with Nvidia Jetson boards. This gives us a ton ... ## NVIDIA Jetson Nano - Docker optimized Linux Kernel DevFeed: [NVIDIA Jetson Nano - Docker optimized Linux Kernel]() Original publisher: [Read original article]() Published: 2019-05-04T01:57:21Z Content type: tutorial Language: en Sources: [Hypriot]() Topics: [Jetson](), [Linux Kernel](), [Docker](), [k3s](), [Kubernetes](), [networking](), [Ubuntu]() Tags: [build-system](), [developer](), [docker](), [jetson](), [k3s](), [kubernetes](), [linux-kernel](), [networking](), [robotics](), [ubuntu]() ### AI overview This tutorial explains how to customize and build a Linux kernel directly on the NVIDIA Jetson Nano DevKit. It focuses on adding kernel settings, including IPVLAN, needed for networking with Docker Swarm, Kubernetes, and k3s, and covers preparing the build environment and downloading the kernel sources. ### Source excerpt Despite the fact that the NVIDIA Jetson Nano DevKit comes with Docker Engine preinstalled and you can run containers just out-of-the-box on this great AI and Robotics enabled board, there are still some important kernel settings missing to run Docker Swarm mode, Kubernetes or k3s correctly. So, let's try to fix this... Analyzing the Linux Kernel In my last blogpost Verify your Linux Kernel for Container Compatibility, I already shared all the details how you can easily verify the Linux kernel for all Container related kernel settings. So, this first part of analyzing the capabilities of the stock Linux kernel 4.9.x provided by NVIDIA is already done and documented. And this was an easy task as well, so everyone who's interested in these details can repeat the task at his/her own device. Let's recap what we did found. Especially there is one important setting which will be used for networking. This feature called "IPVLAN", is required for Docker Swarm mode and it's also used for networking in Kubernetes and k3s. Building your own Linux Kernel Anyway, even when we'd like to include or change only a single kernel setting, we have to customize the kernel configuration and have to compile and build our own Linux kernel. This is typically a common task for a desktop Linux system, but can be pretty ugly and cumbersome if you have to build the kernel for an Embedded Device. When we look back to all the other NVIDIA Jetson boards, like the TK1, TX1 and TX2, this requires a second Linux machine, running Ubuntu 14.04 or 16.04 on an Intel CPU. Then setting up a complete build system for cross-compiling and all these stuff. Honestly, this is a well-known approach for an Embedded Developer, but the good thing now for the Jetson Nano DevKit this is not required any more. Here the good news: you can customize and build your own Linux kernel directly on the Jetson Nano DevKit! You only need an internet connection and some time to perform all steps on your own. BTW, and this is another ## NVIDIA Jetson Nano - Upgrade Docker Engine DevFeed: [NVIDIA Jetson Nano - Upgrade Docker Engine]() Original publisher: [Read original article]() Published: 2019-04-22T06:44:52Z Content type: tutorial Language: en Sources: [Hypriot]() Topics: [Docker](), [Jetson](), [upgrade](), [Ubuntu](), [Development]() Tags: [docker](), [jetson](), [nvidia](), [tutorial](), [ubuntu](), [upgrade]() ### AI overview A tutorial explains how to upgrade the Docker Engine on the NVIDIA Jetson Nano Developer Kit. It starts with Docker 18.06.1-CE and describes using Ubuntu's apt package manager to upgrade to Docker 18.09.2 from the Ubuntu repository. ### Source excerpt In our last blogposts about the NVIDIA Jetson Nano Developer Kit - Introduction and NVIDIA Jetson Nano - Install Docker Compose we digged into the brand-new NVIDIA Jetson Nano Developer Kit and we know, that Docker 18.06.1-CE is already installed, but... But, this isn't the latest available version of the Docker Engine. So, I'd like to point you to a few different options on how to upgrade the Docker Engine to the very latest available version for the NVIDIA Jetson Nano. Check the current Docker Version For this tutorial I'm starting again with a freshly flashed SD card image. Flashing from macOS just takes a few minutes with the Hypriot flash utility, which can be found here https://github.com/hypriot/flash. $ time flash --device /dev/disk2 jetson-nano-sd-r32.1-2019-03-18.img Is /dev/disk2 correct? y Unmounting /dev/disk2 ... Unmount of all volumes on disk2 was successful Unmount of all volumes on disk2 was successful Flashing jetson-nano-sd-r32.1-2019-03-18.img to /dev/rdisk2 ... 12.0GiB 0:03:40 [55.8MiB/s] [=======================================================================================>] 100% 0+196608 records in 0+196608 records out 12884901888 bytes transferred in 220.275160 secs (58494575 bytes/sec) Mounting Disk real 3m47.866s user 0m1.648s sys 0m30.921s Next we have to attach a computer monitor via HDMI cable, mouse and keyboard and of course an Ethernet cable in order to get an internet connect. Now connecting to a micro-USB power supply and follow the instruction on the screen to perform the initial setup of the NVIDIA Jetson Nano Developer Kit. This will take around 5 to 10 minutes and we do have a new Ubuntu 18.04 desktop running on that nice 64bit ARM Cortex-A57 developer board. Pro-Tip on using Docker client: As we have seen in the last blogpost, we have to use sudo when we're calling a docker command in the shell. This can be easily resolved, we only have to issue the following command ones. pirate@jetson-nano:~$ sudo usermod -aG docker pirate Ne ## NVIDIA Jetson Nano - Install Docker Compose DevFeed: [NVIDIA Jetson Nano - Install Docker Compose]() Original publisher: [Read original article]() Published: 2019-04-20T07:51:52Z Content type: tutorial Language: en Sources: [Hypriot]() Topics: [Docker Compose](), [Jetson](), [Python](), [Ubuntu](), [Docker]() Tags: [apt](), [curl](), [docker-compose](), [install](), [jetson](), [nvidia](), [python](), [ssh](), [terminal](), [ubuntu]() ### AI overview This tutorial explains how to install Docker Compose on an NVIDIA Jetson Nano ARM 64-bit board when no direct Docker Compose binary is available. It verifies Python, installs pip and required build dependencies, and installs Docker Compose 1.24.0 via Python pip. ### Source excerpt In our last blogpost NVIDIA Jetson Nano Developer Kit - Introduction we digged into the brand-new NVIDIA Jetson Nano Developer Kit and we did found out, that Docker 18.06.1-CE is already pre-installed on this great ARM board. Today, I want to share some more details on how you can easily install Docker Compose on the Jetson Nano. Install Docker Compose Sadly, there is no binary of Docker Compose available we could install directly on an ARM 64bit board like the NVIDIA Jetson Nano. If we are looking at the official GitHub release page for Docker Compose https://github.com/docker/compose/releases/tag/1.24.0 there are only binaries provided for Intel x86-64 based operating systems like Linux, macOS and Windows - but nothing for ARM 32bit or 64bit systems. This isn't too bad, because Docker Compose is based upon Python and maybe it's easier to install if Python is already available on our board. So, let's check this out the easy way. Docker Compose is based upon Python First, check out if Python is already available on the Jetson Nano. As we have a complete Ubuntu 18.04 LTS desktop operation system this is very likely. But let's verify it directly on a shell running on the Nano. This can be done from the desktop terminal app or via a SSH shell. pirate@jetson-nano:~$ python --version Python 2.7.15rc1 OK, this looks pretty promising. We do have the Python version 2.7.15rc1 already installed on the Nano. Let's try to install Docker Compose via Python PIP. So, first we need to install Python PIP itself. # first, we have to install Python PIP $ sudo apt-get update -y $ sudo apt-get install -y curl $ curl -sSL https://bootstrap.pypa.io/get-pip.py | sudo python # verify the installed Python PIP version pirate@jetson-nano:~$ pip --version pip 19.0.3 from /usr/local/lib/python2.7/dist-packages/pip (python 2.7) Now, let's try to install Docker Compose via Python PIP. Here we're going to pin the latest version of Docker Compose, which is 1.24.0 of the time of writing this blogpost ## NVIDIA Jetson Nano Developer Kit - Introduction DevFeed: [NVIDIA Jetson Nano Developer Kit - Introduction]() Original publisher: [Read original article]() Published: 2019-04-19T12:16:25Z Content type: tutorial Language: en Sources: [Hypriot]() Topics: [Jetson](), [Nvidia](), [Arm](), [configuration](), [Docker]() Tags: [developer](), [devkit](), [docker](), [flashing](), [github](), [hdmi](), [jetson](), [macos](), [nvidia](), [robotics]() ### AI overview An introduction to the NVIDIA Jetson Nano Developer Kit covering its hardware, required accessories, SD card flashing, and initial configuration. The article recommends using a 32 or 64 GB SD card because the compressed image is about 5 GB and the uncompressed image is about 12 GB. ### Source excerpt Let me introduce the brand new NVIDIA Jetson Nano Developer Kit, which is basically a quad-core 64bit ARM Cortex-A57 CPU with 128 GPU cores - suitable for all kinds of maker ideas: AI, Robotics, and of course for running Docker Containers... Unboxing Let's unbox the board and do the initial configuration... When opening the box we'll find the board itself and a short getting started flyer which heads us to the NVIDIA website https://NVIDIA.com/JetsonNano-Start where we can find all basic information about the new DevKit. Please note, there are no other parts included in the DevKit. So, in order to use the DevKit you'll need some additional accessories: MicroSD card (at least 16GByte) Monitor with HDMI or DisplayPort connector HDMI or DisplayPort cable to connect to the monitor USB Keyboard and Mouse Micro-USB power supply (5V, 2Amps for 10 watts) Getting Started Following the official documentation Getting Started With Jetson Nano Developer Kit we have to download and flash the SD card. I don't want to repeat these basic steps here, because it's covered in great detail directy in the official NVIDIA tutorial Write Image to the microSD Card. When downloading the SD card image, please be aware this one isn't small at all. The download is a zipped image file of 5 GByte of compressed data. $ ls -alh jetson-nano-sd-r32.1-2019-03-18.zip -rw-r--r--@ 1 dieter staff 5.3G Mar 15 22:49 jetson-nano-sd-r32.1-2019-03-18.zip The SD card image is a huge 12 GByte data blob. I would strongly recommend to use a SD card size with at least 32 GByte, with the recommended 16 GByte minimum you won't get really happy. So, please do yourself a favour and get a 32 or 64 GByte SD card. $ ls -alh jetson-nano-sd-r32.1-2019-03-18.img -rwxr-xr-x 1 dieter staff 12G Mar 15 19:19 jetson-nano-sd-r32.1-2019-03-18.img For flashing the SD card image you could also use the Hypriot flash tool, which can be found here https://github.com/hypriot/flash. That's exactly what I did from macOS. $ flash --device /dev/