AR on-demand: A practicable solution for augmented reality on low-end handheld devices

This paper presents a practicable approach for augmented reality on embedded systems like PDAs and smart phones. It describes an implementation of a complete AR-assistance system for industrial maintenance. The goal is to provide a future handbook which supports multiple kinds of media documents like SMIL, video and 3D augmented reality views of complex maintenance tasks. The main issue of the system is its usability, as well as its capability to support a mobile worker without overstraining him. The idea is to start from the application and consider AR as a technology that can be invoked on-demand. Similarly, it is assumed offering AR capability on still image snapshots of the real scene is sufficient for many maintenance applications. In particular, two approaches for automatic scene calibration are used, a robust gradient-based marker calibration and a markerless approach, which relies on the registration of a 3D line-model onto the image.