The broader impact of this Small Business Innovation Research (SBIR) Phase I project is a novel tool to visualize indoor spaces. Emerging “property technology” (PropTech) plays an increasing role in the development of new large commercial buildings; it is anticipated that such PropTech will save billions of dollars per year and stimulate remodeling projects that might not otherwise even be undertaken, once it is made more broadly accessible, and may thus be adopted for the remodeling of small businesses, restaurants, and residences. This project develops a simple, rapid, and affordable generation of accurate floor plans of existing structures, leveraging commercial off-the-shelf (COTS) camera technology (implemented in modern tablets and smartphones) coupled with advanced AI software and algorithms that model and identify common features (walls, doors, windows, furniture) in the 3D images generated by such devices. The floor plans so generated will provide an accurate, easily generated framework upon which substantial renovations can then be more easily designed. This is anticipated largely replace the labor-intensive and error-prone manual processes of floor plan generation, thereby reducing operating costs and stimulating new remodeling projects in various markets. This Small Business Innovation Research (SBIR) Phase I project will develop an AI-driven app that quickly and autonomously reconstructs digital spatial layouts of indoor spaces based on walk through calibrated image capture with modern smartphones or tablets, leveraging time-of-flight or structured-light imaging systems together with embedded motion sensors and advanced data analysis. The key intellectual merit of this project is the deep neural networks and optimization algorithms, together with advanced user-interface/user-experience (UI/UX) to provide an intuitive human interface to this data analysis engine. This enables adjustment of the reconstruction rules and visualization of the generated model. The technical hurdles include accurate 3D scene reconstruction and 2D floor plan generation based on sensor-generated point clouds and feature recognition in cluttered environments, and the synthesis of multi-room maps from several single-room models. Rules can be specified by the user as appropriate (defining room adjacency, wall thickness, orthogonality assumptions, etc.). This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
