Revolution in Digital Twin Creation: CHOrD Enables Collision-Free and Organized 3D Indoor Models

The development of digital twins of interiors has made significant progress in recent years. A new method called CHOrD now promises a further leap forward by enabling the creation of house-sized, collision-free, and hierarchically structured 3D models. This technology opens up new possibilities for architecture, interior design, virtual tours, and much more.

CHOrD's Innovative Approach

Traditional methods for generating scene layouts often rely on direct synthesis as a scene graph or object list. CHOrD, on the other hand, uses an innovative 2D image-based intermediate representation of the layout. This approach enables effective collision avoidance by detecting collisions as Out-of-Distribution (OOD) scenarios during generation. This results in more realistic and accurate models.

Controllable Floor Plans and Optimal Layouts

Another advantage of CHOrD lies in its ability to generate scene layouts that conform to complex floor plans with multimodal controls. This allows for the creation of coherent, house-wide layouts that are robust against geometric and semantic variations in room structures. For example, different furniture arrangements within a given floor plan can be simulated and optimized.

A New Dataset for Improved Quality

The developers of CHOrD have also created a new dataset with expanded coverage of household objects and room configurations. This dataset is characterized by significantly improved data quality and contributes to the higher accuracy and detail of the generated models.

Convincing Results and Diverse Applications

Initial tests show that CHOrD achieves state-of-the-art performance on various datasets compared to existing methods. The generated 3D interior models are photorealistic, spatially coherent, and can be adapted to any floor plan variations. The technology has the potential to fundamentally change the way we plan, design, and experience interiors. Applications range from virtual real estate tours to planning escape routes and optimizing room acoustics.

Future Developments and Outlook

The development of CHOrD is an important step towards more efficient and precise creation of digital twins of interiors. Future research could focus on further improving scalability and integrating dynamic elements such as light and shadow. The combination of CHOrD with other AI technologies, such as automatic object recognition, also opens up exciting perspectives for the future of digital twin technology.

Bibliographie: OpenReview.net. (n.d.). *Forum*. Retrieved from https://openreview.net/forum?id=Yj6IdXSOZk Su, C., et al. (2025). *CHOrD: Generation of Collision-Free, House-Scale, and Organized Digital Twins for 3D Indoor Scenes with Controllable Floor Plans and Optimal Layouts*. arXiv preprint arXiv:2503.11958. Hordijk, R. (2019). *Black-Box Optimization with Expensive Function Evaluations*. University of Groningen. Mason, W. H. (2016). *Multidisciplinary system design optimization (MSDO): A Practical and Comprehensive Methodology*. American Institute of Aeronautics and Astronautics. Bommasani, R., et al. (2023). *On the Opportunities and Risks of Foundation Models*. arXiv preprint arXiv:2108.07258. ECAADe. (2023). *eCAADe 2023 Conference Proceedings*. ISARC. (2024). *ISARC 2024 Full Proceedings*. NRF. (2025). *NRF European Research Council (ERC) Book of Abstracts*. New York University. (n.d.). *Computer Science Department Technical Reports*. Retrieved from https://cs.nyu.edu/dynamic/reports/?type=PhD&year=all