AI Integration

As digital transformation accelerates in infrastructure and urban development, the demand for VR simulations and digital twins is rapidly increasing. In response to these trends, we have integrated AI features into UC-win/Road Ver.19.These two approaches—AI chatbots and AI agents—extensively support the creation and maintenance of VR spatial models.

F8 AI® UC Support
The AI chatbot responds to daily operational queries in natural language, covering everything from basic operations and model creation to data settings and simulation conditions. This feature eliminates the need to consult manuals, providing users with the necessary information instantly without disrupting their workflow.

F8 AI® Design Support Generates VR Spatial Models
Simply enter a text prompt, and the AI will generate and update VR spatial models while referencing internal data of UC-win/Road. For example, users can try commands such as the following:

• Add walkways and street trees around intersections
• Update the road alignment using the latest data
• Install lighting for nighttime simulation
• Create a disaster traffic control model

AI assistance enables efficient and faster VR creation without the need for specialized skills, with supported features expanding in future updates.

Another highlight of Version 19 is its compatibility with the Model Context Protocol (MCP). MCP is a universal standard for connecting AI with external systems and data services, enabling AI to function as an agent that retrieves and processes data. Functioning as an MCP server, UC-win/Road connects directly with AI clients while also integrating with external MCP servers provided by national and local governments.

The image below shows an example of AI-generated VR content.

As data services in the transportation and infrastructure sectors adopt MCP, it will unlock new possibilities for AI-driven VR data creation. Specifically, AI will be able to automatically utilize road, urban spatial, map, and transportation data to update VR models. Starting with this AI integration, we look forward to working with our users to unlock new possibilities for infrastructure digital transformation across a wide range of fields, including urban planning, road design, and disaster prevention simulation.

Support for Passthrough View on VR Devices

Previously, we offered the Quest Rift plugin for integration with Meta's Quest series head-mounted displays, as well as various dedicated plugins for other devices. In Version 19, we have transitioned our integration to the OpenXR standard to better support new features in VR devices. While previously restricted to the Quest series, the product now supports devices compatible with SteamVR, such as the HTC VIVE series. Accordingly, the Quest plugin was renamed to the VR Headset plugin.

As a new feature, we have added a passthrough function that overlays 3D models onto the head-mounted display's camera feed, enabling an Augmented Reality (AR) experience. Occlusion settings have been added to the 3D model options, allowing the designated occlusion areas to display the real-world passthrough feed. This feature is available on devices that support the OpenXR passthrough extension, which currently includes the HTC VIVE XR Elite.

Road Editing in 3D View

Previously, creating and editing roads was done from the 2D plan view, but this can now be performed directly in the 3D view. Users can work directly in the 3D space while checking road shapes in real time, enabling more intuitive and efficient road design and modification. The key features are outlined below:

Road creation and editing
Mouse and keyboard operations enable users to create new roads and edit existing ones directly in the 3D view. When creating a new road, users can define the horizontal alignment by sequentially adding points onto the terrain or existing models. Previews are updated in real time with each addition, allowing users to work while checking the positional relationship with the surrounding environment.

During road editing, users can edit eight elements on road models: turning points, vertical points, cross section points, transitions, tunnels, bridges, non-editable terrains, and cross sections. Modifying the positions and shapes of each element is intuitive, and precise value-based input is also supported. Cross-section editing allows for both the direct manipulation of coordinates for constituent points—such as carriageway edges, walkway curbs, and slopes—and the detailed configuration of parameters including materials, widths, and heights. 道路編集中は、道路モデル上に、方向変化点・縦断変化点・断面変化点・トランジション・トンネル・橋梁・非編集地形・断面の8要素を編集できます。

Users can freely pan, zoom, and rotate the 3D view during editing, allowing them to verify the results from any angle. The changes are applied to the 3D model by regenerating the road, allowing users to seamlessly complete both road editing and model inspection within the same 3D view.

Point cloud and 3D model integration and practical Use Cases
This 3D view road editing function delivers maximum effectiveness when combined with point cloud data. By directly editing road shapes while displaying point cloud data in the 3D view, users can intuitively create and modify highly accurate road models based on measured data. It enables real-time verification and editing of the spatial relationships between longitudinal profiles and cross sections while overlaying them with point cloud data.

This function is also highly effective for rapid modification and verification during the road design review stage. For example, by dynamically adjusting walkway widths and curb positions on urban roads, users can instantly check the spatial clearance of surrounding buildings and street trees in 3D. Examining cross-sectional changes—such as altering the number of lanes or adding right-turn lanes in areas near intersections—can be seamlessly verified while editing cross sections and transitions directly within the 3D view. This feedback loop, fully integrated in the 3D space, further enhances both design quality and execution speed.

Expanded Ramp Connecting

In Ver. 19, road connection features have been significantly enhanced. The following are the main additions:

• Ramp connection to existing roads
• Direct connection between roads
• Multi-road connection per lane
• Disconnection of roads

This enhancement enables the branching of different roads per lane from a single connection point. This allows users to express complex branch structures with a simple configuration, making road network creation more flexible and efficient.

In addition, ramps can now be added to or disconnected from existing roads, making it easier to review and partially modify the road network. Since users can flexibly alter the configuration while utilizing existing road data, it minimizes workload and design rework.

Other Improvement

Improved PLATEAU (CityGML file) import
The PLATEAU (CityGML file) import feature supports i-UR 3.2 of the i-UR technical specifications (draft) used in PLATEAU. This allows the import of features defined in i-UR 3.2 from the FY2025 PLATEAU CityGML dataset, which was created based on the Standard Specifications Ver. 5.0. Users can now import urban planning information feature types defined in i-UR into UC-win/Road Ver.19. Since the PLATEAU Standard Product Specification no longer uses 2D coordinates, the previous height settings for 2D data have been removed. Instead, a new feature has been added to vertically offset 3D data when importing CityGML files.

The attribute display feature has been updated to show more attribute types than before.

Support for integration with VISSIM 2025 and 2026
The VISSIM plugin, which integrates with PTV Vision's traffic analysis software, now supports VISSIM 2025 and 2026. In addition to visualizing traffic conditions, the integration with VISSIM allows you to simulate interactions with the driver's vehicle in UC-win/Road when used with a driving simulator.

Improved compatibility of glTF file normal maps
For the glTF file import feature introduced in Ver. 18, we have further improved the visual fidelity of 3D models in Ver. 19. Since the compatibility of normal maps—texture information that expresses surface bumps and textures—has been improved, a wider range of glTF files can now be displayed with their intended appearance. The system utilizes the file's normal map data, and if it is missing, automatically generates it using the industry-standard MikkTSpace algorithm to prevent rendering artifacts.

Similar improvements have been applied to road and terrain rendering, ensuring a more natural appearance at close range.

PBR support for 360-degree rendering and a custom shader plugin
PBR support has been extended to 360-degree stereo rendering and a custom shader plugin that dynamically alter 3D scenes based on model attributes, enabling users to achieve sophisticated visual effects with highly realistic material rendering.

Improvements to the walking algorithm
Version 19 introduces an upgraded VR walking algorithm that analyzes obstacle contours in real time, allowing the simulation to seamlessly slide along walls upon collision. This ensures smooth, human-like movement, enhancing both immersion and operability within VR experiences even in complex environments such as construction sites and plants. This makes it highly valuable for safety education and evacuation drill simulations.