Autodesk, Inc. has secured a major milestone in the Real Estate Development industry with a newly patented machine learning workflow. This innovation focuses on their recent patent, titled ‘Synthetic data generation for machine learning tasks on floor plan drawings’. The patent describes a method and system that provide the ability to generate and use synthetic data to extract elements from a floor plan drawing.

Swanson Reed’s Patent of the Month: February 2026

We are thrilled to recognize this innovation as Swanson Reed’s Patent of the Month for February 2026. It earned this accolade because it represents an outstanding invention in the Real Estate Development sector, revolutionizing the speed and accuracy with which complex architectural floor plans are digitized and translated into functional models.

Patent Abstract

A method and system provide the ability to generate and use synthetic data to extract elements from a floor plan drawing. A room layout is generated. Room descriptions are used to generate and place synthetic instances of symbol elements in each room. A floor plan drawing is obtained and pre-processed to determine a drawing area. Based on the synthetic data symbols in the floor plan drawing are detected. Orientations of the detected symbols are also detected. Based on the detected symbols and orientations, building information model (BIM) elements are fetched and placed in the floor plan drawing.

Meeting the U.S. R&D Tax Credit Rules

To qualify for the Research and Development (R&D) tax credit in the United States, an innovation must satisfy the IRS Four-Part Test. Autodesk’s system meets these stringent rules through the following demonstrations of technical innovation:

• Permitted Purpose: The research was undertaken to create a new, improved, and highly reliable software function—automating the extraction of 2D floor plan symbols and converting them to BIM elements.
• Technological in Nature: The development process fundamentally relies on the hard sciences, specifically advanced computer science, computer vision, and machine learning principles.
• Elimination of Uncertainty: At the outset, there was technical uncertainty regarding the optimal capability and methodology required to accurately train an AI to detect symbols and spatial orientations across highly variable, non-standardized floor plan drawings.
• Process of Experimentation: Autodesk engineers engaged in a systematic trial-and-error process, evaluating different synthetic data placement strategies, algorithmic pre-processing techniques, and ML architectures to successfully detect and map BIM elements.

3 Practical Applications Qualifying for R&D Tax Credits

Companies integrating or building upon this patented technology could also claim R&D tax credits if they are developing custom applications such as:

1. Automated 2D to 3D BIM Conversion Software: Developing a proprietary software tool that leverages this synthetic data methodology to automatically generate 3D Building Information Models from legacy 2D floor plan PDFs. The systematic testing of machine learning algorithms to ensure highly accurate symbol extraction across various regional architectural styles would qualify as experimental R&D.
2. Automated Code Compliance and Safety Checking Systems: Engineering a new compliance system that uses synthetic data generation to train models to detect missing safety elements (like fire exits, ADA clearances, or structural supports) in floor plans. Evaluating and iterating upon the symbol detection algorithms to reach a high confidence threshold successfully eliminates technical uncertainty.
3. Smart Facilities Management Integration: Creating a dynamic application that parses raw architectural layouts to automatically map physical assets (HVAC, electrical components, plumbing) directly into facilities management dashboards. Formulating and testing the algorithms required to orient, fetch, and integrate the correct BIM elements dynamically involves rigorous technological experimentation.