ABSTRACT

Many augmented reality (AR) systems are developed for entertainment, but AR and particularly mobile AR potentially have more application possibilities in other fields. For example, in civil engineering or city planning, AR could be used in combination with CityGML building models to enhance some typical workflows in planning, execution and operation processes. A concrete example is the geo-referenced on-site visualization of planned buildings or building parts, to simplify planning processes and optimize the communication between the participating decision-makers. One of the main challenges for the visualization lies in the pose tracking, i.e. the real-time estimation of the translation and rotation of the mobile device to align the virtual objects with reality. In this paper, we introduce a proof-of-concept fine-grained mobile AR CityGML-based pose tracking system aimed at the mentioned applications. The system estimates poses by combining 3D CityGML data with information derived from 2D camera images and an inertial measurement unit and is fully self-sufficient and operates without external infrastructure. The results of our evaluation show that CityGML and low-cost off-the-shelf mobile devices, such as smartphones, already provide performant and accurate mobile pose tracking for AR in civil engineering and city planning.

摘要

开发了许多用于娱乐的增强现实（AR）系统，但是AR，尤其是移动AR在其他领域有更多的应用可能性。例如，在土木工程或城市规划中，可以将AR与CityGML构建模型结合使用，以增强规划，执行和运营流程中的一些典型工作流程。一个具体的例子是对规划中的建筑物或建筑物部件进行地理参考的现场可视化，以简化规划过程并优化参与决策者之间的沟通。可视化的主要挑战之一在于姿势跟踪，即对移动设备的平移和旋转进行实时估计以使虚拟对象与现实对齐。在本文中，我们针对上述应用引入了概念验证的基于移动AR CityGML的细粒度姿势跟踪系统。该系统通过将3D CityGML数据与从2D摄像机图像和惯性测量单元获得的信息相结合来估计姿态，并且完全自给自足，无需外部基础架构即可运行。我们的评估结果表明，CityGML和低成本现成的移动设备（例如智能手机）已经为土木工程和城市规划中的AR提供了高性能和准确的移动姿态跟踪。