Build and learn
September 18, 2023

Viam & NVIDIA: Raising the bar on smart machine ML

Sara Gharnit
Senior Content Marketer
On this page

Imagine a world where  deploying your machine learning (ML) model to your smart machine is as quick as brewing a cup of coffee—just 5 minutes. This isn’t a dream, it’s just using our Viam ML Model Service.

This month, we partnered with NVIDIA to spotlight our cutting-edge ML Model Service: an integrated solution that makes training and deploying ML models on your smart machines faster and easier than ever before. Here’s why and how:

ML's potential to boost real-time smart machine performance

The data landscape is undergoing a rapid expansion. As we generate and consume an unprecedented volume of data, leading organizations are dramatically transforming their customer experiences, maximizing efficiency and significantly reducing costs.

However, the smart device sector tells a different story. Compared to other industries, their pace of ML adoption is noticeably slower due to challenges ranging from the safety concerns due to the dynamic environments smart machines operate in, to hardware limitations and a lack of standardization. 

Additionally, financial constraints, expertise gaps, and longer development cycles further complicate swift machine learning integration, limiting the ability of developers to apply the latest in software innovation and improve overall hardware performance.

The overarching question: How can innovators in the smart machine space harness, interpret, and leverage the data they capture?

Our answer: Make creating and deploying custom ML models to a machine far quicker and simpler. 

Empowering your smart machines with ML

Viam bridges the gap between getting data off the machines and being able to use it to make the machine perform better with the insights ML provides. Within months of Viam ML’s debut, forward-thinking businesses and developers have already harnessed Viam ML to turbocharge their devices. 

Here’s why they’re turning to our platform:

  • Adaptive Intelligence: Keep your machines on the forefront of innovation by introducing cutting-edge models whenever the need arises.
  • Effortless Integration: Get your pre-trained models on board in a snap and utilize data to amplify your machine's capabilities.
  • Streamlined Data Flow: Whether it’s instantaneous data captures by sensors or backlogged uploads, Viam’s cloud platform organizes and processes it all.
  • One-Stop Automation: Take your data, annotate it with labels, gear up the ML models, and roll them out to your entire fleet, without breaking a sweat.
  • Modular Flexibility: Due to Viam’s modularity, you can use the ML Model Service with other Viam features or by itself. Build on top of your pre-existing architecture to enhance your machine’s capabilities without disturbing its core setup.

And on top of it all, Viam transforms the traditional, time-intensive data processes. Extracting and training data on machines, once a weeks-long endeavor, is now a swift, streamlined task. 

Learn more about the many features of Viam ML and how it can make your smart machines even smarter.

Viam & NVIDIA: Elevating ML together

Software-wise, we set the gold standard, developing our solutions to be in sync with any computer. This includes those by some of the most trusted and popular in the business, including NVIDIA. Their Jetson Orin Nano, packed with formidable GPUs, supercharges processing—facilitating faster and more efficient ML and AI tasks.

Integrating the Jetson Orin Nano with the Viam app? As shown in the video below, it's a breeze. Users simply:

  • SSH into the board.
  • Download and install the viam-server.
  • Define the board and select both component and model types. Coding skills? Not required.

An added benefit? Viam's Modular Registry now includes the Viam Triton Inference Server Module for Jetpack, streamlining its integration with your smart machine.

For those unfamiliar, the Registry is a space for creating, sharing, and utilizing custom drivers, making it more adaptable for all smart machines. This ensures that users can make the most of the Orin device's capabilities, optimizing resources and allowing the CPU to efficiently handle additional robot tasks.

At Viam, we believe in the power of collaboration. Together with NVIDIA, we're paving a straightforward and practical path for businesses and developers to tap into the potential of ML for smart machines. Our partnership simplifies the process, making ML more accessible and efficient.

Curious about the real-world impact of our combined efforts? Dive into our comprehensive webinar, check out our detailed documentation on our ML Model Service, and try out our ML features within the app for yourself. 

We can’t wait to see what you build with it!

twitter iconfacebook iconlinkedin iconreddit icon

Find us at our next event

May 6, 2025
May 6, 2025
,
07:00-09:00 PM EST

Elastic New York Meetup

In Person
New York, NY
Monitor and automate the physical world with Elastic and Viam. Join us for a demonstration of gathering data from a fleet of sensors, visualizing it with Kibana, and creating alerting rules that trigger in real life.
Secure your spot
May 5, 2025
May 7, 2025
,

Shift Miami

In Person
Perez Art Museum 1103 Biscayne Blvd, Miami, FL
Interested in robotics, but don't know where to start? Meet Viam in Miami, where Adrienne Tacke will discuss how to get up and running, even if you're "just" a software developer.
Join Us
May 7, 2025
,

Deploying and scaling AI with hardware

Virtual
Curious how startups are using Viam to build smart, vision-enabled products, even on low-power hardware? Join Viam engineers for a live computer vision demo and Q&A.
Join Us
Jun 12, 2025
Jun 16, 2025
,

JS Nation

In Person
Amsterdam
WebRTC is most often associated with building video and text chat into browsers but this peer-to-peer technology can also be used to monitor and control machines from anywhere in the world! Join Nick Hehr to learn about industrial arms, DIY rovers, and dashboards of data in real time.
Register Now
Jun 23, 2025
Jun 25, 2025
,

Open Source Summit North America 2025

In Person
Denver, CO
Edge-based computer vision gives us real-time insights, but getting that data where it needs to go without high bandwidth, lag, or hardware strain is a big challenge. Learn how to build a fast, event-driven vision pipeline.
Learn More