Seeed Studio at GTC 2024, Pioneering the Future of AI at the Edge

GTC 2024 emerged as a significant highlight for Seeed!  As NVIDIA Robotics Elite Partner, we took pride in hosting a booth at the “AI at the Edge” Pavilion, where we displayed our cutting-edge reThings series, driven by NVIDIA’s Robotics Jetson Orin. During GTC’s opening day, we hosted a Partner Reception, extending a warm welcome to NVIDIA teams, our edge AI partners, customers, and community. We’re sincerely thankful for everyone’s support and enthusiasm. It’s a privilege to support you as the AI hardware partner!

At the show, we demonstrated that by uniting our ultra-low-power microcontrollers with Jetson Orin, we have masterfully blended TinyML and Local Large Language Models.  This endeavor showcases our dedication to pushing the envelope in AI technology at the edge, emphasizing a seamless fusion of innovation and efficiency

Top trends we’ve observed throughout #GTC2024

1. The buzz around Generative AI was undeniable. From groundbreaking software to cutting-edge hardware, the push into Large Language Models and their application across various fields was a focal point for many.

2.  Multimodal AI is on the rise, marking a significant advancement from basic image recognition to generating valuable content. NVIDIA’s display of multimodal vision language agents running on Orin was particularly notable. In line with this trend, our integration of TinyML with LLMs aims to seamlessly connect complex physical world data to the digital realm.

3.  Robotics is big:  NVIDIA Jetson powering a wide range of robots. The “special guests,” Orange and Green BD-X robots, powered by Jetson Orin, received a warm welcome during Jensen Huang’s keynote speech at GTC.

4.  Another concern among enterprises is deployment. NVIDIA’s NIM’s release underscores the industry’s need for easier deployment of advanced foundation models.

5.  Omniverse, industrial metaverse, and digital twins were shown to offer valuable insights across various sectors for industrial use cases.

Bring Generative AI to the Edge

We are excited to share the GTC experience featuring Dustin Franklin sharing his insights on integrating generative AI models on Jetson Orin devices, including LLMs, VLMs, and EfficientViT(vision models for efficient high-resolution vision, created by MIT HAN Lab). There are new Small Language Models (SLMs) updated on Gen AI lab! SLMs represent a growing class of language models that have <7B parameters – for example, StableLM, Phi-2, and Gemma-2B. Their smaller memory footprint and faster performance make them good candidates for deploying on reComputer J3011, powered by Orin Nano.

Elaine (Edge AI partnership manager at Seeed) discussed with Dustin Franklin(Principal Engineer at NVIDIA) on how to bring LLMs to Jetson Orin by following the Gen AI Lab

We united the new capabilities of TinyML with Gen AI Lab’s Large Language Models deployment, we are presenting an advanced AI system that extends from the edge to the cloud.  

  • TinyML comes to life through ultra-low-power sensors, including cameras and devices that capture environmental and auditory data.  These sensors feed streamlined, processed data directly into our LLMs.  
  • Running on edge devices powered by NVIDIA Jetson Orin, the LLMs provide sophisticated, context-aware insights and real-time responses, enhancing scene perception with layers of complexity.  
  • AWS underpins this system, providing the necessary infrastructure for deployment, management, and scalability.  In parallel, Node-RED provides the cohesive interface that weaves together disparate components and services, forging a cohesive network of devices.  

This network is not just smart;  it is a self-optimizing symphony of technology, where devices and AI are in constant dialogue, making informed decisions independently — a testament to the autonomous synergy of a truly advanced technological ecosystem. 

Jensen Huang made an instant connection to AWS’s vision of Gen AI & IoT Chess Robot, powered by AWS Bedrock and Seeed‘s Jetson devices

Leveraging Generative AI, robotic arms can independently strategize and engage with a smart chessboard, exemplifying the seamless integration of robotic control with AI and cognitive capabilities.

Our reComputer J4012, powered by the Jetson Orin NX 16GB, is designed for advanced AI capabilities, handling chess strategies, evaluations, and decision-making. Coupled with the reServer, these systems control robotic units capable of exhibiting human-like reasoning and intricate decision patterns during gameplay. In a notable 5-minute demonstration featuring Jensen Huang and Channa Samynathan(Senior Worldwide Technical Lead at AWS), the precision and future scalability of AWS’s demo was showcased, especially with the potential integration of new GenAI models on Amazon Bedrock.

AWS team demonstrated the demo to Jensen Huang

Deliver edge AI into industries, working with the powerful ecosystem 

The event turned into a grand reunion with the community and our partners!   We were delighted to witness our reThings hardware of the Jetson series featured across many partner booths, contributing to field demonstrations of vision AI, robotics, and generative AI at GTC.

  • Arrow introduced a proof of concept (PoC) for an Autonomous Mobile Robot (AMR) utilizing NVIDIA Jetson AGX Orin, featuring integrations with RPLIDAR, motor drive, power management, and in-vehicle networking tailored for industrial environments.
  • Edge Impulse released a new suite of tools developed on Nvidia’s Omniverse and AI platforms, aimed at bringing advanced AI models to a wider range of Seeed’s edge devices, including MCUs, Linux SBCs, and the Jetson Orin series.
  • Nota AI utilized reServer J4012 in their Intelligent Transportation Systems for enhancing accident prevention and enabling rapid response.
  • Numurus offers NEPI, a comprehensive platform for Jetson-compatible robotic automation, featuring sensor and control drivers, AI and automation applications, ROS support, system management tools, and customizable interfaces.
  • OpenZeka showcased the integration of video analytics and LLM capabilities on their Cordatus AI platform, compatible with all Seeed Jetson Orin series.
  • Roboflow presented live demos of their latest releases, including YOLO-World for zero-shot object detection, leveraging the power of an NVIDIA Jetson Orin for edge-based execution.

ICYMI: GTC On-demand sessions of Vision AI, Generative AI, and Robotics:

  • S62315 Robotics and the Role of AI: Past, Present, and Future
  • S62980 Fireside Chat With Percy Liang and Jim Fan: The Future of Foundation Models
  • S62409 Empowering Collaborative Robots: The Future of AI Vision with Digital Twins  
  • S63374 Breathing Life into Disney’s Robotic Characters with Deep Reinforcement Learning
  • S63034 Panel Discussion on the Impact of Generative AI on Robotics
  • SE62675 Meet Jetson, the Platform for Edge AI and Robotics
  • SE62934 Introduction to AI-Based Robot Development With Isaac ROS
  • SE62933 Overview of Jetson Software and Developer Tools
  • SE62935 Getting Started: Generative AI at the Edge

Next Livestream of MonthlychatCEOLive #003 

That’s just a glimpse at our experiences from GTC 2024.  Keep an eye out as we delve deeper into these discoveries and more, sharing reflections on our participation and the future trajectory of these groundbreaking technologies.   What’s next?   Innovation doesn’t pause post-GTC.  We’re excited to invite you to our next livestream, MonthlychatCEOLive #003.

🚀 Explore Innovation at GTC 2024: A Symphony of TinyML and Local LLMs.

🕹 Set a reminder on YouTube

In our next session, focused on the topic of “minimizing the costs of large language models,” we’ll showcase our engaging demo from GTC.   We had a fantastic time at the conference and are already looking forward to next year.  

A big thank you to NVIDIA for hosting us and to everyone who visited our booth.  Get in touch anytime with our team to discuss your edge AI needs!

Seeed NVIDIA Jetson Ecosystem

Seeed is an Elite partner for edge AI in the NVIDIA Partner Network. Explore more carrier boards, full system devices, customization services, use cases, and developer tools on Seeed’s NVIDIA Jetson ecosystem page.

Stands at the forefront of global hardware innovation, we commit to making technology accessible to all. Specializing in IoT and AI, Seeed offers a spectrum of open hardware, design, and agile manufacturing aimed at lowering the threshold of hardware innovation.

About Author

Leave a Reply

Your email address will not be published. Required fields are marked *


March 2024