This Small Business Innovation Research (SBIR) Phase I project proposes to automate the creation of digital 3D building models and 3D street maps used in the personal navigation, architecture, engineering, defense and homeland security markets. It currently requires hundreds of hours to create a 3D model of an average city building or streetscape using a tripod-mounted or vehicle-mounted laser scanner and available CAD software. This research will potentially reduce that time significantly. Automated 3D modeling has been a primary goal of the CAD industry for a generation. The problem can be divided into two separate challenges: 1) identifying and extracting observed surfaces, and 2) subsequently extending those surfaces automatically to form a solid 3D model. The proposed research involves developing algorithms to extrapolate observed building surfaces and extend them to correctly intersect other surfaces, completing the solid 3D model entirely automatically.
The potential to create highly accurate 3D building models and 3D street maps in minutes will enable cost-effective modeling of entire cities. Accurate 3D city models made widely available to the public has the potential to have a profound impact on the mapping and personal navigation markets as well as commercial architecture, engineering projects, virtual tourism and first responder effectiveness. In the Architecture, Engineering and Construction (AEC) market alone, firms spend nearly $800 million per year on manually creating 3D building models. The proposed effort will not replace the existing CAD programs; rather, it will work with CAD programs to remove the tedious and expensive manual steps of model creation from vehicle-mounted and tripod-mounted laser scanners. If successfully developed, the commercial potential of automated high-definition 3D building models and 3D street maps is significant.
This Small Business Innovation Research Phase I project developed several critical components of a software platform that will create fully 3D computer models of buildings, streetscapes and entire cities. 3D models are used extensively in the architecture/engineering (A/E), personal navigation/mapping, and government markets. However, it currently takes hundreds of hours and tedious manual tracing of laser scan or photogrammetry point clouds to create a full 3D model of an average city building. The ultimate output of this Phase I research, EdgeWise City Mapper software, is a fundamental change from current CAD technology and will reduce 3D modeling time by 99%. The intellectual challenge of this research is profound—automating 3D model creation is a massive mathematical undertaking, one that has vexed the CAD/mapping community for decades. Algorithmically accounting for the innumerable geometries of building facades and automatically creating a light-weight, accurate, fully 3D streetscape model has never before been seriously attempted, according to all published literature. The main goal of this Phase I research was to create a technology framework that will solve the occlusion problem: areas of building facades that are hidden or occluded from laser scan view and comprise the bulk of the work to manually complete 3D building models. The first step in solving this problem was to create a novel database architecture to account for and define the fundamental components of façade geometry. The company developed a model database and created a series of topological rules that allowed its software to infer much about the occluded regions of 3D scanned objects. At the end, the company was able to successfully generate, store, and manipulate multiple models using this database structure. Once this database architecture was developed and successfully tested, the next task was to devise a method to automatically extend observed surfaces through occluded regions and correctly connect those surfaces to each other in order to form the complete 3D building model. The company created a series of unique algorithms to force approximately coplanar surfaces to line up exactly so that they could ultimately form topologically correct joins. Next, the company developed a novel approach to merging overlapping polygons such as would be collected from a mobile scanner. Finally, the automatic modeling algorithms were tested with several synthetic datasets. The final results exceeded expectations. The algorithms were able to automatically and correctly reconstruct the entire 3D solid objects given only partially observed surfaces. All of the major components of the Phase I research were completed successfully, and the state-of-the-art for full 3D automated modeling took a significant step forward. The company developed a solid foundation on which to build fully automated 3D modeling software. The broader impacts of this research are far reaching and potentially transformative. Software that can create near-instantaneous, highly accurate 3D building models will enable cost-effective modeling of entire cities. The widespread availability of fully 3D city models and streetscapes will have a profound impact on the personal navigation, commercial real estate, design/construction, first responder, security and defense industries. Although this project was high-risk due to the enormity of the algorithmic challenge, it was high reward – with a commercial potential close to one billion dollars. In the Architecture/Engineering market alone, firms spend nearly $800 million per year on manually creating 3D building models. Successful commercialization of this research will reduce those costs by up to 99%. In the highly competitive personal navigation/mapping market, firms such as Google, Nokia, and Microsoft have unsuccessfully tried to apply their 2.5D aerial mapping technology to street-level mapping, which requires a completely different technological approach. ClearEdge anticipates these mapping providers will become the company’s largest customers. EdgeWise City Mapper will deliver a completely innovative technology and a totally new economic equation to the marketplace—one that will spur the widespread adoption of high-resolution 3D building models and 3D streetscapes among consumers and industry alike.
