Comparison of Tracking Methods for Augmented Reality-Supported Damage Mapping of Grid Shell Structures

As powerful mobile devices become available and robust tracking methods are further developed, Augmented Reality (AR) becomes increasingly suitable for use on construction sites. For the renovation of the Mannheim Multihalle, the world's largest free-form timber grid shell structure, various AR tracking methods are examined for their suitability for carrying out damage mapping: manual initialization (four points or vector input) and localization by cloud anchors. As the uniform grid of the grid shell makes orientation difficult during manual damage mapping, using a smartphone AR application to correctly recognize the laths and nodes is promising. Three key test criteria were used to compare the methods: the correct assignment of the laths, the smallest computationally recognized error and sufficient stability, meaning the absence of drift. In addition, the effort required for preparation and handling during use was considered. For the use case of damage mapping using a smartphone, the tests showed that although manual initialization using vector input is the most suitable method, even this method is not satisfactory. Therefore, in the outlook, measures to facilitate optical tracking and the use of differential GPS and synchronization with point cloud data are discussed.