![](/images/stories/topics/artificialintelligence.gif)
Wayne Arvidson, Dell Technologies, and Charbel Aoun, NVIDIA, explore how AI-enabled technologies and digital twins can transform urban management, driving cities toward a more sustainable future.
Creating digital twins starts with understanding the core concept and the tools available
We’re all aware of the challenges cities face regarding sustainability, needing to simultaneously reduce their carbon footprints, optimise resources, and minimise waste production. With AI-enabled solutions now more widely available to cities and local authorities at lower costs than ever before, its ability to support cities in solving these challenges has come to the fore.
AI-enabled smart parking can help dramatically reduce the amount of CO2 created by idling vehicles by more efficiently guiding drivers to free spaces. Similarly, dynamic traffic routing that prioritises public transport can reduce a fleet’s fuel usage significantly. Waste management is another area where AI can make an impact, utilising waste levels with sensors and generating a dynamic route for waste collection, as opposed to a fixed route which wastes time and resource.
Our latest video interview goes into more detail on these examples, and the scope for AI usage in sustainability and climate action is far wider than this – take NVIDIA’s involvement in Project Earth-2 as an example. It uses AI and high-powered computing to simulate the planet, allowing climate scientists to collaborate on managing disasters such as city flooding and hurricanes. AI tools can predict weather, track icebergs, and identify pollution, which are vital for environmental management.
At a local level, though, for those on the ground working for cities, what are the base-level requirements and considerations?
Getting your hands on the right data
When you’re in a city, data comes from everywhere, so it’s important to know what you need to serve your purpose. Existing CRM systems, databases, and documents within the city are essential starting points for making the most of what you have with AI. With the advent of multimodal generative AI and large language models (LLMs), we can manipulate, learn from, and generate value from the wide range of data that already exists.
The concept of multimodality allows us to access and analyse text, video, audio, and documents. For example, a city could prompt an LLM to identify dirty streets based on predefined criteria. This kind of capability, previously only possible by spending a lot of money on a solution, is now more accessible thanks to LLMs.
Sensors are also important sources of sustainability data – cameras and air quality sensors, for example, providing valuable audio and video feeds. The word "sensor" relates not only to traditional devices like cameras or air quality sensors, though, but now extends even to social media – a sensor, in its own way, of human feedback. For instance, if people are venting on social media about traffic, crowdsourcing opinions from thousands of people is a form of sensing that provides valuable data. Innovative approaches, combined with traditional technologies, support cities in managing their dynamic needs.
One thing to remember across this conversation for city IT teams is the operational areas that have traditionally operated in silos. These areas contain valuable data that can be integrated into broader AI initiatives. When approaching data integration for AI, we have to start with the problem we aim to address and the outcome we want. For instance, if we want to assess traffic impact and associated pollution and its impact on personal experience, we need to consider a number of data sources; traffic lights, camera data for vehicle counts, social media sentiment data, pollution sensor data, and emergency services data related to respiratory issues. LLMs make integrating such diverse data sources more feasible than ever before.
Using LLMs, we can create tracks that connect all these data sources, enabling multimodal analysis. This allows for deep learning and meaningful insights from diverse data sets. We can query an AI model, much like we interact with chatbots, and get comprehensive insights based on data from multiple sources.
Thinking about the supply chain
Focusing on climate change shouldn’t lead to the deployment of infrastructure that consumes more power or requires more cooling.
When deploying extensive technology infrastructure across a city, it’s crucial to ingest once and receive many insights. The Dell AI Factory with NVIDIA embodies this idea. It provides a single infrastructure that can serve multiple use cases, eliminating technological silos throughout a city. This approach, coupled with energy-efficient engineering, ensures that we can derive valuable insights while minimising power consumption and cooling requirements.
Both Dell Technologies and NVIDIA are deeply committed to sustainability. Dell, for instance, is one of the world’s largest recyclers of technology, with over 95 per cent of its packaging materials currently recycled. Dell also uses more than 90 per cent recycled material in its technology production.
NVIDIA technology supports not only the most efficient of the Green 500 supercomputers, but they also support seven of the top 10 Green 500, highlighting its commitment to climate and efficiency – something that is embedded in the company’s DNA and product roadmap. The recently-launched Blackwell GPU is generally 20 times more energy efficient than traditional GPUs, making a significant impact on high-performance computing and data centres.
Predicting and adapting to sustainability challenges
With the constant changing of conditions in cities, increasingly related to the changing climate, its no longer enough for cities to monitor and react; they must be proactive, predicting and projecting outcomes. Digital twins are becoming more and more feasible solutions for cities here, but their creation must be outcome-led.
Creating digital twins starts with understanding the core concept and the tools available. Generative AI, for instance, can be instrumental in building digital twins, using platforms like NVIDIA’s Omniverse. Often, when people think of generative AI in a city context, they think first of digital assistants. However, its application in developing digital twins offers immense value by transforming raw data into comprehensive virtual models, and then learning from that data to provide further synthetic data to conduct modelling in the longer term.
Cities should begin by identifying what they aim to simulate. Simulation isn’t a new concept, but achieving high fidelity results – accurately reflecting the real world in both visualisation and physics – is vital when time is of the essence, as it is when considering climate risks. Traditional simulations often fall short because they don’t adhere to the laws of physics or lack realistic environmental interactions. A platform like Omniverse, however, offers high-fidelity simulations that abide by these laws and provides a realistic environment.
For example, if a city wants to simulate weather conditions like flooding, they need a wealth of data from various sources: weather stations, sensors, historical data, and real-time updates. This data encompasses rain, wind, sun, and more.
The goal is to create a simulation environment that closely mimics reality, incorporating all these variables. An engine capable of mimicking the physical world at high fidelity while considering all these variables allows cities to build accurate models – models that aren’t academic exercises, but dynamic simulations that combine mathematical precision with real-world visuals.
The journey toward sustainable urban development starts with leveraging existing data and integrating advanced technologies like AI and digital twins. Cities need to identify key areas of focus, gather comprehensive data, and use high-fidelity simulation platforms to create accurate and actionable models. By doing so, they can anticipate and mitigate the impacts of climate change, optimise resource use, and enhance overall quality of life for their residents.
---
Autor(en)/Author(s): Wayne Arvidson and Charbel Aoun
Quelle/Source: Smart Cities World, 09.07.2024