Nvidia and partners bring physical AI to cities and industrial infrastructure

AI Video & Visuals


Physical AI is becoming the foundation of smart cities, facilities and industrial processes around the world.

Nvidia works with companies like Accenture, Avathon, Belden, Deephow, Milestone Systems and Telit Cinterion to enhance operations around the world with physical AI-based recognition and inference.

The ongoing loop of simulation, training and deployment of physical AI provides sophisticated industrial automation capabilities, making cities and infrastructure safer, smarter and more efficient.

For example, physical AI applications can automate tasks that are potentially dangerous to workers, such as heavy machinery work. Physical AI can improve transportation services and public safety, and can also detect defective products in factories, among others.

The need for this is greater than ever. The numbers tell the story:

Infographic statistics: Low quality and manufacturing flaws have lost $7 trillion a year. 2.8 million workers die each year from industrial accidents and labor-related illnesses. 514,000 industrial robots installed around the world in 2024. It spent $300 billion a year on the EU's public order and security. By 2030, we have predicted a global labor shortage of 50 million people.

The infrastructure that allows you to perceive perception, reason, and behavior relies on video sensors and modern vision AI capabilities. Simplify the development, deployment, and scaling of video analytics AI agents and services from edge to cloud using the NVIDIA Metropolis platform.

Below we present five major companies that promote physical AI and five important Nvidia Metropolis updates today at the Siggraph Computer Graphics Conference, enabling such advancements.

Five companies that advance physical AI

Global Professional Services Company Accenture Working with Belden, a leading provider of complete connectivity solutions, we are increasing the safety of our workers by creating smart virtual fences that allow factories to place around large robots to prevent accidents with human operators.

Smart fence images.
Image courtesy of Accenture and Belden.

Smart Virtual Fence is a physical AI safety system that uses OpenUSD-based digital twins and physically grounded simulations to model complex industrial environments. Using computer vision-based mapping and 3D spatial intelligence, the system adapts to increased variability in dynamic human robot interactions that occur in modern ShopFloor environments.

Accenture taps the Nvidia Omniverse platform and Metropolis to build and simulate these smart fences. With Omniverse, Accenture created a digital twin of robotic arms and a worker moving into space. Also, along with Metropolis, the company trained AI models and deployed it to the edge with video intake and real-time inference capabilities of the NVIDIA Deepstream Software Development Kit (SDK).

AvattonIndustrial Automation Platform Providers use Nvidia's blueprints in Video Search and Summary (VSS) which is part of Nvidia Metropolis to provide manufacturing and energy facilities with real-time insights that improve operational efficiency and worker safety.

Reliance British Petroleum Mobility Limited, a leader in the fuel and mobility sector in India, has increased productivity by using Avathon Video Intelligence products during the construction of gas stations by increasing the productivity by lowering safety compliance standards, lowering unsafe violation cases and saving thousands of working hours.

Deep How We have developed a “Smart Know-How Companion” for employees in manufacturing and other industries. Companions use Metropolis vss Blueprint to convert key workflows into bite-sized multilingual video and digital instructions to improve onboarding, safety and floor operator efficiency.

Faced with the need for high-end skills, Anheuser-Busch Inbev, a retired company of skilled workers, turned to the Deephow platform to transform standard operating procedures into easy-to-understand visual guides. This reduced onboarding time by 80%, improved training consistency, and improved employee knowledge retention over the long term.

Milestone SystemIt provides one of the world's largest platforms for managing IP video sensor data in complex industrial and urban deployments, and creates the world's largest library of real-world computer vision data through its platform Project Hafnia. Among its features, the platform is provided to physical AI developers who have access to customized Vision Language Models (VLMs).

The milestone system that taps Nvidia nemo curator helped to build a fine-tuned VLM for intelligent transport systems for use in VSS blueprints and develop AI agents that better manage urban roads. Milestone Systems is also considering using the new open and customizable NVIDIA COSMOS reason VLM for physical AI.

Internet company Terrylion The Nvidia Tao Toolkit 6 was integrated into an AI-powered visual inspection platform that supports multimodal AI and provides high-performance inference, using Vision Foundation models such as Foundation Poses alongside other Nvidia models. TAO brings low-code AI capabilities to the Telit platform, enabling manufacturers to quickly develop and deploy accurate, custom AI models for defect detection and quality control.

Updates to 5 Nvidia Metropolis for Physical AI

Major updates to Nvidia Metropolis enhance developers' capabilities to build physical AI applications faster and easier.

Cosmos Reason VLM

Latest Version of COSMOS Reason – NVIDIA's highly open, customizable, 7 billion parameter inference VLM physical AI VLM allows context video understanding, temporal event inference for metropolis use cases. Its compact size makes it easy to deploy from the edge to the cloud, making it ideal for automating traffic monitoring, public safety, visual inspections and intelligent decisions.

VSS Blueprint 2.4

VSS 2.4 allows you to quickly augment existing vision AI applications for COSMOS reasons and provide powerful new features to your smart infrastructure. The enhanced set of BluePrint application programming interfaces gives users the flexibility to select the capabilities to select specific VSS components and features to enhance computer vision pipelines with the generated AI.

New Vision Foundation Model

The NVIDIA TAO Toolkit includes a new suite of new vision foundation models, along with advanced fine-tuning methods, self-monitoring learning, and knowledge distillation capabilities to optimize deployment of physical AI solutions across edge and cloud environments. The NVIDIA Deepstream SDK includes a new inference builder that allows for seamless deployment of TAO 6 models.

Companies around the world, including Advex AI, Instrumental AI and Spingence, experiment with these new models and Nvidia Tao to build intelligent solutions that optimize industrial operations and drive efficiency.

Nvidia Isaac Sim Extension

New extensions to the NVIDIA ISAAC SIM Reference Application help to solve common challenges in vision AI development, such as limited labeled data and rare edge case scenarios. These tools simulate human-robot interactions, generate rich object detection datasets, create incident-based scene and image caption pairs to train VLMSs, accelerate development and improve AI performance in real terms.

Extended hardware support

All of these Metropolis components can now run on the NVIDIA RTX Pro 6000 Blackwell GPU, the NVIDIA DGX Spark desktop supercomputer, and the NVIDIA Jetson Thor platform for physical AI and humanoid robotics.

Cosmos Reason 1 and Nvidia Tao 6.0 are now available for download. Sign up to vss 2.4, cosmos reason VLM tweak update, and alert you when nvidia deepstream 8.0 becomes available.

Please take a look Siggraph's Nvidia Research Special Onders And by joining Nvidia at the conference, which runs through Thursday, August 14th, we'll learn more about how graphics and simulation innovations can drive industrial digitalization.

look Let me know Regarding software product information.



Source link

Leave a Reply

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