STTR Phase I: Rapid and Efficient Scene Modeling for Law Enforcement and Disaster Response

This Small Business Technology Transfer Research (STTR) Phase I project will produce technology capable of constructing three-dimensional (3D) models of scenes accurately, rapidly, and at low cost. The technology leverages recent developments in computer vision and enhances them for operation under varying ambient conditions while increasing speed of model construction, thus enabling practical applications. Key system design requirements identified include: 1) ability to produce complete and accurate high fidelity models quickly; 2) robustness to varying ambient conditions; 3) ease of use by a non-specialist without extensive training; and 4) low cost. The proposed research objectives include: improvements to state-of-the-art algorithms for robust and efficient 3D modeling from images; the development of an approach for dense 3D reconstruction using different sensor modalities; the development of a sensor view point planning algorithm; the creation of techniques for the identification of missing model data; and the development of a system for merging data from different sources into a coherent 3D model of the environment. The work plan includes a thorough assessment of the system?s ability to reconstruct a scene accurately and completely, as well as the benefit of reconstructing scenes using sensors deployed by low-cost vehicles. The broader impact/commercial potential of this project will be the widespread use of 3D modeling for scene documentation. Systems currently available for this application are expensive, slow, bulky, and their use requires special training. The technology resulting from this work will enable the production and commercialization of systems that are faster, more affordable, and easier to use by non-experts, thereby simplifying their adoption by a larger number of law enforcers, prosecutors, insurers, and other government agencies. In the transportation domain, for example, a speedy scene reconstruction is essential to restore the flow of traffic when accidents happen. On busy highways, traffic backup can grow at a rate of up to a mile per minute of delay in clearing the accident site. This currently limits the complete documentation of accidents to only those cases where fatalities occur. The technology proposed will allow the documentation of larger number of incidents. Additionally, a great reduction in the time and complexity involved in determining the cause of accidents is expected. The potential long-term benefits of this reduction include a more rapid evolution of regulation and policy to deal with chronic causes of traffic accidents, which will expedite the implementation of road safety mechanisms.