CRII: CHS: Structurally-Aware Computation for Geometry Acquisition and Design

Dartmouth College, Hanover NH

Investigators

Abstract

Current modeling software is primarily concerned with geometric descriptions of objects, while real-world considerations such as structural stability and robustness have been treated as a separate issue. This separation lacks the insights that structural information can bring to the design and modeling process, and makes the modeling of physical artifacts difficult and inefficient. The PI's goal in this project is to alleviate this problem by establishing a research program in a new field of structural geometry processing, a cross-pollination of digital geometry processing and structural engineering, which will infuse structural information into all stages of geometric modeling, from scanning to reconstruction to editing and design, and which will lead to more efficient 3D models with a higher degree of accuracy and realism that behave in a physically-correct manner. 3D printing and digital fabrication have started a revolution in design and manufacturing; the emergence of desktop 3D printers has delivered widespread accessibility to non-experts for realizing complex objects. This research will allow the development of more robust, reliable, and usable tools for creating digital content. Aside from their direct relevance to computer graphics, the geometry processing technologies to be developed here will also have applicability to cultural heritage applications, opening up new possibilities for archiving and visualizing historic sites, and for analyzing the stability and safety of historic masonry buildings that have survived for centuries. The PI's interdisciplinary approach will also open new areas of research in the respective fields of geometry processing and structural engineering. This research will impact three core thrusts: structurally-informed 3D geometry acquisition; geometric modeling and design of stable structures; and prototyping with 3D fabrication. This first thrust will develop new methods to capture the geometry of large architectural sites by exploiting structural priors. Scans often suffer from missing information due to obstructions and inherent construction limitations. The PI will improve on reconstruction methods from scan data by incorporating constraints for structurally-correct, and hence more faithful, digital representations of buildings. Algorithms will be developed to automatically translate captured surface geometry into volumetric mass models that can be analyzed by structural mechanics methods. The second thrust will develop the foundation for a new class of Computer-Aided-Design (CAD) tools that harmoniously integrate structural objectives. Advances in surface and volumetric modeling have had wide influence in engineering and design. While contemporary buildings now feature an explosion of free-form shapes, these expressive shapes are often achieved at the expense of high material and construction costs. Existing software lacks the ability to aid designers in improving geometry, for example, to reduce internal forces and required material. The PI will investigate shape optimization methods for exploring energy landscapes linked with structural constraints. The third thrust will validate the optimized geometry from the first two thrusts using digital fabrication technology. The process of turning a virtual model into a physical artifact is subject to many constraints. The PI will develop computational tools to address problems in robustness, economic use of material, and printability for large-scale prototypes. Scaled 3D printed models will enable effective physical validation of the modeling concepts previously discussed.