SBIR Phase I: Automatic Reconstruction of As-is Building Information Model from Indoor Point Cloud Data for Planning Purposes

Award ID: 1942348

The broader impact of this Small Business Innovation Research (SBIR) Phase I project will develop a new platform automating the 3D model generation process from cloud data. Several companies have developed tools to facilitate the modeling process, but despite the benefits offered by their tools the process is still semi-automatic, expensive, time-consuming, labor-intensive, error-prone, and requires a designer. The proposed project will create 3D solid representations of structural, architectural, and mechanical components (e.g., beam, ceiling, column, floor, pipe, wall); openings (e.g., door, window); and furniture (e.g., sofa, bed, chair, table). Furthermore, the platform will allow use of the generated model for visualization, coordination, and scene management; simplify planning tasks; and improve communication and collaboration. The proposed solution will enable users with little or no experience designing 3D models to have the capability to increase productivity, reduce planning time and cost, and increase collaboration. This Small Business Innovation Research (SBIR) Phase I project aims to develop a platform that provides a quick, easy, and economical solution to generate 3D solid representations of indoor scenes. The platform will use artificial intelligence (AI) to extract the geometric and semantic information embedded in point cloud data and fit a solid geometry for generating the 3D solid representation of indoor scenes such as living areas, offices, utility rooms, and mechanical rooms. The design of the proposed platform will consist of three major objectives: (1) automatic generation of 3D solid models, (3) new tools to use the model for planning purposes, and (3) integration with industry-standard communication tools. The first task has three principal research objectives: (a) multi-scale feature extraction, (b) semantic identification of the point cloud main elements through machine learning algorithms, and (c) 3D solid model generation using the elements' primitives and attributes. 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.

NSF Program Director: Peter Atherton