Toolkits to make AI based simulations of cityscapes with Unreal Engine.

Madara Premawardhana
3 min readMar 17, 2023

--

Unstyled UE5 cityscape of London city

Unreal Engine has been a popular game engine used by developers to create visually stunning and immersive games. However, it’s also finding a growing application in other industries, such as architecture and city planning. The engine’s ability to simulate cityscapes and landscapes has made it an ideal platform for developers to create virtual worlds for architects and planners to test their designs before construction even begins.

One of the most interesting and powerful applications of Unreal Engine is its use in simulating cityscapes. With the help of AI and machine learning algorithms, the engine can generate incredibly realistic and detailed virtual cities. Developers can create a variety of environments, from dense urban areas to sprawling suburban landscapes, complete with a range of buildings, roads, and other infrastructure.

One example of Unreal Engine’s use in cityscape simulations is the software created by the start-up ‘Improbable.’ Their software, called ‘SpatialOS,’ uses Unreal Engine to create massive multiplayer online games that are capable of supporting thousands of players. However, it’s also being used in other industries, such as architecture and city planning.

Link to SpatialOS GitHub: https://github.com/spatialos

SpatialOS can simulate entire cities, complete with traffic patterns, weather, and even crowds of people moving about the city. Architects and planners can use this technology to create virtual models of their designs and test them in a virtual environment. They can then make changes and adjustments to the design before any actual construction begins, potentially saving both time and money.

Another example of Unreal Engine’s use in cityscape simulations is the ‘Urban Simulator.’ Developed by the University of North Carolina, this software uses Unreal Engine to create realistic virtual cities. The simulator can be used by urban planners and architects to test their designs in a virtual environment, helping them to identify potential problems and issues before construction begins.

Link to Urban Simulator: https://github.com/UDST/urbansim

The Urban Simulator also incorporates AI algorithms to help simulate the behavior of virtual pedestrians and vehicles. This allows architects and planners to test the impact of their designs on traffic patterns and pedestrian flow, helping them to create more efficient and livable cities.

Cesium is an attempt to integrate the whole world data in one global simulation which allows the buffering depending LOD(Level Of Distance) on zoom levels. Following are methods in which Cesium allows users to integrate itself with AI simulations.

  • Real-world data integration: The Cesium plugin allows for real-world data integration, which can be helpful in AI simulations. For example, if you are simulating a city and you want to incorporate real-world data about the terrain, buildings, and roads, the Cesium plugin can help you do that. This can make your simulations more realistic and accurate, which can help with AI training.
  • High-performance graphics: The Cesium plugin is designed for high-performance graphics, which can be important for AI simulations. AI simulations often involve complex environments and large amounts of data, and the Cesium plugin can help you render that data quickly and efficiently. This can be particularly important if you are using AI to control objects in real time.
  • Collaboration: The Cesium plugin allows for collaboration between teams working on the same project. This can be helpful in AI simulations, where multiple teams may be working on different aspects of the simulation (e.g., graphics, physics, AI). The Cesium plugin can help you coordinate those efforts and ensure that everyone is working with the same data and information.

Link to Cesium: https://github.com/CesiumGS/cesium-unreal

--

--

Madara Premawardhana

PhD Student at the University of Buckingham, School of Computing