NVIDIA Expands Physical AI in Cities and Industry with New Partnerships

Lawrence Jengar
Aug 11, 2025 16:34

NVIDIA collaborates with leading companies to advance physical AI, enhancing safety and efficiency in urban and industrial environments through innovative technologies and platforms.

NVIDIA is pushing the boundaries of physical AI, enhancing safety and efficiency in urban and industrial environments through strategic collaborations, according to NVIDIA’s blog. The technology giant has teamed up with several prominent companies, including Accenture, Avathon, Belden, DeepHow, Milestone Systems, and Telit Cinterion, to integrate physical AI capabilities into various operational processes worldwide.

Innovations in Physical AI

Physical AI is revolutionizing the way cities and industries function by automating complex tasks and improving safety measures. NVIDIA’s partnerships aim to leverage AI-based perception and reasoning to create safer and more efficient environments. The continuous cycle of simulating, training, and deploying physical AI is enhancing industrial automation capabilities, making infrastructures smarter and more responsive.

For instance, physical AI applications can automate hazardous tasks, such as managing heavy machinery, thereby reducing risks for workers. Additionally, these technologies are improving public safety and transportation services, along with identifying defective products in manufacturing processes.

Key Collaborations and Developments

Accenture, in collaboration with Belden, is working on creating smart virtual fences around large robots in factories to prevent accidents. This initiative uses NVIDIA’s Omniverse platform to simulate digital twins and enhance safety measures through computer vision-based mapping and 3D spatial intelligence.

Avathon utilizes NVIDIA’s video search and summarization blueprint to provide real-time insights that boost operational efficiency and worker safety in manufacturing and energy facilities. This has proven beneficial for companies like Reliance British Petroleum Mobility Limited in India, which has seen improvements in safety compliance and productivity.

DeepHow is transforming industrial training by developing a “Smart Know-How Companion” that converts workflows into multilingual video guides, improving onboarding and knowledge retention. Anheuser-Busch InBev has already benefited by reducing onboarding time and enhancing training consistency.

Milestone Systems is creating a vast computer vision data library to aid in the development of AI agents for intelligent transportation systems. This initiative is supported by NVIDIA’s NeMo Curator and Cosmos Reason VLM, facilitating better management of city roadways.

Telit Cinterion has integrated NVIDIA’s TAO Toolkit into its AI-powered visual inspection platform, enabling manufacturers to develop accurate AI models for defect detection and quality control.

Advancements in NVIDIA Metropolis

NVIDIA has announced several updates to its Metropolis platform to streamline the development of physical AI applications. These include the introduction of the Cosmos Reason VLM for contextual video understanding and temporal event reasoning, making it suitable for various applications from traffic monitoring to public safety.

The VSS Blueprint 2.4 update offers enhanced flexibility for augmenting vision AI applications, while new vision foundation models in the TAO Toolkit optimize deployment across different environments. Additionally, NVIDIA Isaac Sim extensions address challenges in vision AI development by simulating interactions and generating datasets for training AI models.

These innovations are supported by NVIDIA’s advanced hardware, including the RTX PRO 6000 Blackwell GPUs and the DGX Spark desktop supercomputer, ensuring seamless development and deployment from edge to cloud.

Image source: Shutterstock