Building High-Performance Edge AI Solutions with Nvidia Jetson and Edge Impulse

author avatar

22 Apr, 2024

NVIDIA Jetson AGX Orin Module and Developer Kit

NVIDIA Jetson AGX Orin Module and Developer Kit

Nvidia Jetson and Edge Impulse facilitate Edge AI development, offering computational power and model training platforms for real-world applications.


The emergence of Edge AI is revolutionizing data processing, enabling decisions to be made right where information is gathered. Nvidia's Jetson Orin series offers powerful computer modules specifically designed for edge computing tasks. This article explores the seamless integration between Jetson Orin devices and Edge Impulse, a platform for building machine learning models.

By explaining the features of the Nvidia Jetson products and Edge Impulse's user-friendly model building tools, this article equips readers with the knowledge to create high-performance edge AI applications. We'll explore the capabilities and applications of these technologies across various industries, highlighting the potential of Edge AI to transform embedded systems and reshape our daily lives.

The Nvidia Jetson Orin Series

The Nvidia Jetson series is engineered for advanced computing in various domains, including robotics, embedded devices, and edge AI applications. This series aims to provide high computing performance for AI-enabled tasks. It is structured to support developers and businesses in implementing AI applications with greater efficiency. The lineup comprises the AGX Orin, Orin NX, and Orin Nano models, each designed for specific performance requirements and use cases.[1]

Nvidia Jetson AGX Orin

The Nvidia Jetson AGX Orin stands as the flagship device of the Orin series, designed for the most demanding AI applications that require high computational power and efficiency. With up to 275 TOPS (Tera Operations Per Second) of AI performance, it surpasses its predecessor, the AGX Xavier, marking a significant advancement in AI capabilities. This module is built on Nvidia's Ampere architecture, featuring an integrated GPU with 2048 CUDA cores and 64 Tensor Cores that are optimized for AI acceleration.

Further reading: Understanding Nvidia CUDA Cores: A Comprehensive Guide

AGX Orin is equipped with an 8-core ARM Cortex-A78AE CPU, providing robust processing capabilities for complex computational tasks. It also includes 32GB of LPDDR5 memory and 64GB eMMC storage, offering ample space and speed for demanding applications. The AGX Orin is ideal for autonomous machines, high-resolution sensor processing, and embedded AI computing, providing the computational horsepower necessary for real-time processing and decision-making.

Nvidia Jetson Orin NX

The Nvidia Jetson Orin NX is the mid-range offering in the Orin series, designed to offer a balance between performance and power efficiency. It is targeted at developers and enterprises needing a compact, yet powerful AI computing solution. The Orin NX delivers up to 100 TOPS of AI performance, making it well-suited for edge AI applications, intelligent video analytics, and robotics.

The Orin NX is built on the same Ampere architecture as the AGX Orin and features a 6-core ARM Cortex-A78AE CPU, alongside 896 CUDA cores and 28 Tensor Cores for AI acceleration. It comes with 16GB of LPDDR5 memory and offers options for 8GB or 16GB eMMC storage, providing flexibility for various application needs. The Orin NX strikes an excellent balance between size, power consumption, and performance, catering to applications where space and energy efficiency are critical.

Nvidia Jetson Orin Nano

The Nvidia Jetson Orin Nano is the entry-level model in the Orin series, designed for developers and enterprises looking for an efficient and cost-effective AI computing solution. It provides up to 40 TOPS of AI performance, making it suitable for entry-level edge AI applications, smart cameras, and low-power IoT devices.

Fig. 1: Nvidia Orin Nano

The Orin Nano utilizes a 4-core ARM Cortex-A78AE CPU and includes 512 CUDA cores and 16 Tensor Cores for AI acceleration. It offers 8GB of LPDDR5 memory and comes with options for 8GB or 16GB eMMC storage, aligning with the needs of less demanding applications that still require efficient AI processing capabilities. The Orin Nano is an excellent choice for applications that need to balance cost, power consumption, and AI performance.

This table offers a comparison to help understand the differences in capabilities across the Nvidia Jetson Orin series:


Jetson AGX Orin

Jetson Orin NX

Jetson Orin Nano

AI Performance

Up to 275 TOPS

Up to 100 TOPS

Up to 40 TOPS


Ampere architecture with 2048 CUDA cores and 64 Tensor Cores

Ampere architecture with 896 CUDA cores and 28 Tensor Cores

Ampere architecture with 512 CUDA cores and 16 Tensor Cores


8-core ARM Cortex-A78AE

6-core ARM Cortex-A78AE

4-core ARM Cortex-A78AE







8GB or 16GB eMMC

8GB or 16GB eMMC

Intended Applications

High-performance applications like autonomous machines, high-resolution sensor processing

Mid-range applications like edge AI, intelligent video analytics

Entry-level edge AI applications, smart cameras, low-power IoT devices

Benefits of using Nvidia Jetson for Edge AI Product Development

Embedded systems powered by Nvidia Jetson Orin offer a versatile foundation for integrating advanced AI capabilities directly into a wide array of products, from industrial machines to consumer electronics, thereby accelerating development cycles and enhancing product functionality. The benefits of utilizing Jetson for embedded systems in product development are manifold.[2]

Flexible, End-to-End Hardware Solutions

Nvidia Jetson Orin devices provide a comprehensive system on module (SOM) solution that includes everything needed for powerful AI computing at the edge—GPU, CPU, memory, power management, and high-speed interfaces. This integrated approach caters to a broad spectrum of power, performance, and form-factor requirements, making Jetson Orin devices adaptable to various project needs. From the compact Orin Nano for entry-level applications to the robust AGX Orin for the most demanding AI tasks, developers have access to a range of options to best suit their project's specific demands.

Advanced Processing with Next-Gen AI Capabilities

Jetson Orin devices are at the forefront of edge AI technology, offering advanced processing capabilities essential for next-generation AI applications. With support for the most comprehensive AI software stack available, including NVIDIA JetPack, Metropolis microservices, production-ready Isaac ROS packages, and application-specific reference AI workflows, developers can seamlessly integrate cutting-edge AI technologies into their products. This eliminates the need for costly internal AI resources and accelerates product development and deployment at scale.

Support for Cloud-Native Technologies and IoT Applications

In an era where billions of IoT edge devices operate autonomously, the ability to efficiently manage and update these devices is crucial. Jetson Orin devices embrace cloud-native technologies and workflows, such as containerization and orchestration, to expedite edge AI product development. This support not only facilitates agile deployment and management at scale but also ensures that products remain up-to-date with the latest AI capabilities throughout their lifecycle.

Streamlined Product Integration for Reduced Time-to-Market

The Jetson ecosystem, comprising hardware partners and software vendors specialized in AI solutions and cloud services, offers an extensive network of resources designed to streamline product integration. With access to peripherals like sensors, cameras, and connectivity modules, as well as vertical-specific AI solutions, developers can leverage these comprehensive solutions to accelerate their time-to-market. Solution integration providers assemble these components into unique combinations, further simplifying the development process and enabling faster deployment of innovative products.

Applications in Different Industries

Nvidia Jetson has been used for transformative solutions across various industries, showcasing the vast potential of embedded systems powered by AI.[3] Let's explore some of the most impactful applications:

Healthcare: Real-Time AI Risk Prediction for Kidney Patients in Taiwan

Taipei Veterans General Hospital utilized the Nvidia Jetson edge AI platform to develop a real-time AI risk prediction model for kidney dialysis patients, addressing Taiwan's high prevalence of kidney dialysis patients. This model predicts heart failure risk with 90% accuracy by analyzing streaming data from dialysis procedures, including patient medical records, test results, and medication information. The implementation of this technology demonstrates how early detection and prompt decision-making can significantly improve patient outcomes.

Automotive: Pioneering Autonomous Special Vehicles Infrastructure

Pegasus Technology, a company building L4 autonomous driving solutions, leveraged Nvidia Jetson AGX Orin to create an infrastructure for autonomous special vehicles, marking a milestone in the evolution of autonomous driving. With over 1000 TOPS of computing power and high redundancy, modularity, and scalability, Nvidia Jetson enabled Pegasus Technology to develop high-performance autonomous control solutions. 

Logistics: Revolutionizing Delivery with Drones

Zipline, a drone delivery startup, completed over 800,000 orders by utilizing Nvidia Jetson-powered drones for autonomous navigation and precision landings. Their service, which now includes delivery for companies like Pagliacci Pizza and GNC, demonstrates the efficiency and environmental benefits of using AI-powered drones for logistics.

Fig. 2: An autonomous bot built by Nvidia on the Jetson platform

Safety: Fighting Wildfires with AI

Chooch, a startup, integrated computer vision and generative AI technologies with the Nvidia Jetson platform to enhance wildfire detection and response. By reducing false positives from 2,000 a week to eight, their solution provides firefighters with real-time alerts, enabling quicker and more efficient wildfire management.

Environment: Protecting the Great Barrier Reef with AI

In collaboration with Google and Australia's Commonwealth Scientific and Industrial Research Organisation (CSIRO), an AI model was developed to monitor the Great Barrier Reef and detect outbreaks of crown-of-thorns starfish (COTS) using Nvidia Jetson AGX Xavier. This real-time monitoring system allows for quicker interventions, showcasing how AI can play a vital role in conserving marine ecosystems.

Further Reading: Spotlight on Innovations in Edge Computing and Machine Learning

These success stories across healthcare, automotive, logistics, safety, and environmental protection illustrate the transformative impact of Nvidia Jetson technology. By enabling real-time AI applications and advanced processing capabilities, Nvidia Jetson is making a way for innovative solutions that address complex challenges in diverse sectors.

Building Edge AI Solutions with Nvidia Jetson Orin and Edge Impulse

NVIDIA Jetson Orin devices and Edge Impulse together create a game-changer for building and deploying AI solutions on the spot. This powerful combo combines high-performance processing with an easy-to-use development platform, making cutting-edge AI accessible to more developers and projects than ever before.

About Edge Impulse

Edge Impulse is an AI model development ecosystem built for edge applications. It offers a comprehensive suite of tools that streamline the process of collecting data, designing, training, and deploying machine learning models directly onto edge devices. This platform is designed to democratize AI, enabling developers to build smarter products without requiring deep expertise in machine learning. Edge Impulse supports a wide range of sensors and data types, making it versatile for various use cases.

Model Building Process

The process of creating effective AI models with Edge Impulse involves several key steps, all facilitated by the platform's user-friendly interface and robust feature set:

  • Data Collection: Edge Impulse allows for the easy ingestion of data from sensors or existing datasets, providing a solid foundation for model training.

  • Data Labeling and Management: Once data is collected, developers can label and manage it within the platform, utilizing tools that help identify patterns or anomalies crucial for training purposes.

  • Feature Extraction: The platform employs signal processing and feature extraction techniques to transform raw data into a format suitable for machine learning models, optimizing performance for edge deployment.

  • Model Training and Validation: Developers can train machine learning models using the processed data, with the platform providing real-time feedback on performance and accuracy. Validation tools ensure that models are ready for real-world deployment.

  • Deployment: Edge Impulse enables the seamless deployment of trained models onto devices, including those powered by Nvidia Jetson Orin, ensuring models run efficiently in their intended environment.

Development Board and Tool Compatibility

The synergy between Edge Impulse and Nvidia Jetson devices is most evident in their compatibility and flexibility in supporting a wide range of projects. Edge Impulse's platform is designed to be compatible with the Jetson series, leveraging its powerful computing capabilities to run complex AI models at the edge. This compatibility ensures that developers can deploy sophisticated AI solutions that require real-time processing, high accuracy, and low latency, directly onto Jetson Orin-powered devices.

The combination of Edge Impulse's streamlined model development process with the high-performance computing of Nvidia Jetson Orin devices opens up new possibilities for edge AI applications. From enhancing industrial automation and improving healthcare diagnostics to enabling smarter consumer devices and advancing environmental monitoring, the potential use cases are vast. By simplifying the development and deployment of AI models, Nvidia Jetson Orin and Edge Impulse are empowering developers to create innovative solutions that push the boundaries of what's possible at the edge.

Check out the official documentation by Edge Impulse for getting started with Nvidia Jetson: NVIDIA Jetson | Edge Impulse Documentation


The future of embedded systems is bright with Edge AI, powered by tools like Nvidia Jetson and Edge Impulse. These technologies will change how we interact with devices and promise to make our lives not only simpler but also significantly better. 

This article explored how Nvidia Jetson and Edge Impulse are driving the Edge AI revolution. Their impact can be felt in various fields, from healthcare to environmental protection. By making AI development easier, they open doors to creating smarter and more efficient solutions.


[1] Nvidia Jetson Modules, Nvidia, [Onlines], Available from:

[2] Embedded Systems for Product Development, Nvidia, [Online], Available from:

[3] Embedded Systems - NVIDIA Jetson Success Stories, Nvidia, [Online], Available from: