New Tools for Reproducible AR Experiments: ARC and ARW

The development and testing of augmented reality (AR) applications is often a challenge, as experiments take place under non-reproducible conditions. To solve this problem, we have developed two innovative tools: ARC (AR Capture) and ARW (AR Workstation). ARC is a native Swift app that uses ARKit to record AR sessions, including RGB-D data, tracking information and lighting properties. This data is stored in a human- and machine-readable format. ARW, a Unity-based tool, allows these recordings to be played back and manipulated on the desktop. This allows AR scenes to be analyzed repeatedly under different conditions. Integration into Unity allows developers to register events, insert their own algorithms and perform reproducible tests, e.g. for geometric and photometric reconstructions. Complex experiments such as voxel generation, ray tracing or the application of neural networks are thus considerably facilitated. Our work shows how ARC and ARW can improve research in AR development. Future updates aim to extend data acquisition and improve fidelity to enable even broader fields of application. A link to the publication can be found here: http://dx.doi.org/10.1109/ISMAR-Adjunct64951.2024.00086