NVIDIA Partners Propel Physical AI and Smart City Innovations in Dublin, Ho Chi Minh City, and Raleigh

With two out of every three people expected to live in urban centers by 2050, adding an estimated 2.5 billion new residents, cities face an urgent demand for sustainable planning and enhanced public services. The smart traffic management market alone is projected to reach $20 billion by 2027, as AI becomes critical for managing growing populations and tourism-driven congestion.

Improved traffic flow is just one aspect where AI is revolutionizing public spaces and urban processes. Many of these groundbreaking physical AI technologies are currently on display at the Smart City Expo World Congress (SCEWC) in Barcelona.

NVIDIA's expanding ecosystem of physical AI partners, encompassing simulation and mapping companies, software vendors, manufacturers, and systems integrators, is showcasing their latest smart space applications. These solutions are built upon the NVIDIA Blueprint for smart city AI, initially announced at NVIDIA GTC Paris in June.

This comprehensive blueprint integrates digital twins with NVIDIA Omniverse libraries, synthetic data generation, and AI model training, including vision language models (VLMs). It also leverages video analytics AI agents powered by the NVIDIA Blueprint for video search and summarization (VSS).

Together, these components create a streamlined workflow, enabling rapid simulation of real-world urban conditions, extensive sensor integration, and real-time vision AI. Recent enhancements to the blueprint include new NVIDIA Cosmos world foundation models and VLMs, which generate photorealistic synthetic data and provide advanced physical reasoning capabilities. The latest version of the VSS blueprint, part of the NVIDIA Metropolis platform, further aids in training and building vision AI agents.

Step-by-step intelligent traffic system workflows and technical recipes are available in new cookbooks for NVIDIA Cosmos Predict, NVIDIA Cosmos Transfer, and NVIDIA Cosmos Reason. These tools empower leading companies to bring secure, scalable physical AI to a growing list of cities.

As urban populations surge and infrastructure ages, challenges like congestion, climate resilience, and safety demand new levels of intelligent responsiveness. Smart city deployments require a seamless integration of next-generation AI, cloud and edge computing, ultrafast networking, and open data platforms. NVIDIA’s partners are crucial in solving these complex technological and real-world challenges.

Mapping and spatial analytics leader Esri is collaborating with NVIDIA to develop an AI agent for the City of Raleigh, North Carolina. This agent ingests, analyzes, and visualizes vast amounts of camera data, generating real-time alerts and insights on an interactive geospatial map.

This system allows city operators to quickly respond to issues, optimize traffic flows, and improve infrastructure design, thereby boosting overall urban mobility. By tapping into the NVIDIA Blueprint for smart city AI, Raleigh can aggregate data to address resident concerns, automate streetlight timing to reduce delays, and decrease carbon emissions from idling vehicles.

Milestone Systems is integrating generative AI into its XProtect video management platform, enabling users to extract valuable analyses and insights from video streams and libraries. This functionality, powered by NVIDIA Cosmos Reason VLMs, allows for automated report generation and alert review.

Milestone post-trained these VLMs with 75,000 hours of compliant traffic video for both European and U.S. contexts. The company plans to offer these specialized VLMs-as-a-service to developers, with general availability expected later this year. These new, GDPR-compliant features can reduce operator alarm fatigue by up to 30%, automating video review and filtering out false alarms. Cities like Dubuque, Iowa, and Genoa, Italy, are set to pilot these advanced video intelligence solutions.

Linker Vision, a pioneer in deploying the full NVIDIA Blueprint for smart city AI, is extending its physical AI solutions to Ho Chi Minh City and Danang in Vietnam. Building on its success in Kaohsiung City, Taiwan, where its vision AI platform reduced incident response times by up to 80%, Linker Vision is now harnessing the blueprint to develop city-scale AI solutions.

These solutions, utilizing AVES Reality’s simulation-ready 3D digital twins with NVIDIA Omniverse, can simulate and monitor traffic and construction activities. This helps ensure projects stay on schedule while significantly improving urban mobility and operational efficiency.

Ireland’s Smart Dublin organization is tackling urban challenges with physical AI, leveraging Cesium, a 3D geospatial platform from Bentley Systems, and NVIDIA Omniverse. This setup enables real-time data visualization and analysis for micromobility—including walking, cycling, and scooter use—across the city.

Additionally, AI transportation technology company VivaCity employs NVIDIA Jetson and Metropolis platforms on its AI-powered computer vision sensors. These sensors provide Dublin with highly accurate, multimodal data and insights on cyclists, motor vehicles, and pedestrians. This data is critical for understanding traffic patterns, road user behavior, and identifying dangerous sites, thereby enhancing road safety and traffic flow throughout the Irish capital.

Smart Dublin is also integrating AI agents into a digital twin for faster insights and alerts to city operators. An initial step involved integrating VivaCity data into a Cesium-powered digital twin to visualize bike route utilization. Interestingly, adding rainfall data revealed a negligible impact of poor weather on cyclist numbers. Future plans include incorporating AI features like natural language search. Furthermore, Bentley’s AI technology Blyncsy utilizes NVIDIA Cosmos and Metropolis to generate synthetic data for analyzing road conditions and improving maintenance.

Deloitte is applying AI to automate street inspections across thousands of crosswalks for city transportation departments. This initiative aims to protect vulnerable roadway users, such as pedestrians and bikers.

Using Cosmos Predict, Deloitte converts static intersection images into photorealistic, physically accurate videos. Cosmos Transfer then generates videos simulating various real-world scenarios, including fog, rain, snow, and low light, bringing each environment to life. Finally, Cosmos Reason helps evaluate these videos to identify potential areas for improvement, making the entire inspection process insightful and dynamic.

A broad ecosystem of hardware and other partners supports these smart city deployments, including AAEON, Advantech, Aetina, Dell Technologies, HPE, OpenZeka, and YUAN High Technologies. Many of these partners are exhibiting physical AI demos running on NVIDIA RTX PRO Servers, NVIDIA DGX Spark, and NVIDIA Jetson Thor modules on the show floor.

Attendees can also explore Akila’s physical AI technologies in the Monaco-Monte-Carlo train stations and its digital twin solutions for the University Mohammed VI Polytechnic campus in Morocco. K2K will demonstrate its real-time analytics solution, which utilizes NVIDIA Cosmos Reason and the VSS blueprint to optimize roadway safety, traffic, and waste management. See city-scale AI and urban innovation in action by joining NVIDIA and its partners at SCEWC, and sign up to be notified when the NVIDIA Blueprint for smart city AI becomes available.