SBIR Phase II: Reconstructing Consistently Detailed City-Scale Environments From Incomplete 2D and 3D Data

这个小企业创新研究（SBIR）第二阶段项目的更广泛的影响/商业潜力是通过完美的虚拟副本更全面地了解现实世界的能力。这个SBIR二期项目使创建真实城市的虚拟复制品成为可能，不仅可以展示世界的样子，还可以完美地再现每个物体和每个细节，让用户与之互动。这些虚拟副本可以快速有效地训练士兵和急救人员，训练自主地面和飞行车辆，将新建筑放置在密集的城市中，模拟灾难性天气的影响，并通过游戏探索想象的场景。为这些和许多其他应用绘制真实世界的地图和3D模型已经花费了数十亿美元，但这些虚拟城市是通过艰苦的手工方法创建的，可能需要几个月到几年的时间，或者缺乏关于世界上什么是必要的信息。这个SBIR二期项目将使快速自动创建现实世界任何区域的详细3D副本成为可能。这个小型企业创新研究（SBIR）第二阶段项目将推进重建高度详细的世界3D模型的最新技术，用于各种商业应用。该项目引入了将原始多模态传感器数据转化为描述世界和沉浸式交互式3D模型的语义信息的新方法。它将通过使用计算机智能而不是人工来解析传感器数据（如照片和激光扫描），从而节省时间、金钱和创建现实世界区域精确3D模型所需的人工努力。由此产生的3D模型和描述世界的底层语义信息将直接用于游戏引擎和模拟软件、分析工具、渲染软件等。这项研究将建立在相关的第一阶段项目的基础上，首先改进可以在现实世界的虚拟副本中识别和复制的细节，然后向付费客户展示商业化的准备。该项目的结果将是针对初始细分市场的市场准备产品，能够在客户可以随时使用的3D模型中准确地再现真实的城市，以及证明客户对产品需求的牵引力。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase II project is the ability to more completely understand the real world through a perfect virtual copy. This SBIR Phase II project makes it possible to create virtual reproductions of real cities that not only show how the world looks, but reproduce every object and every detail perfectly, allowing users to interact with it. These virtual copies make it possible to quickly and effectively train soldiers and first responders, train autonomous ground and flying vehicles, place new construction into a dense city, simulate the effects of catastrophic weather, and explore imaginary scenarios through games. Billions of dollars are already spent mapping and 3D modeling the real world for these and many other applications, but these virtual cities are created through painstaking manual methods that can take months to years, or lack necessary information about what is in the world. This SBIR Phase II project will make it possible to rapidly and automatically create detailed 3D copies of any area of the real world.This Small Business Innovation Research (SBIR) Phase II project will advance the state of the art in reconstructing highly detailed 3D models of the world for diverse commercial applications. This project introduces new methods for turning raw multimodal sensor data into semantic information describing the world and immersive, interactive-ready 3D models. It will remove the time, money, and manual effort necessary to create accurate 3D models of real world areas today, by using computer intelligence instead of human effort to parse sensor data like photographs and laser scans. The resulting 3D models and underlying semantic information describing the world will be used directly in game engines and simulation software, in analysis tools, in rendering software, and beyond. This research will build on the associated Phase I project, first improving the detail that can be identified and reproduced in virtual copies of the real world, then showing readiness for commercialization with paying customers. The result of the project will be a market-ready product for an initial market segment, capable of accurately reproducing real cities in 3D models that customers can readily use, as well as traction that demonstrates customer need for the product.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.