Generating vision based Lego augmented reality training and evaluation systems

Abstract

The creation of training applications using Augmented Reality (AR) is still a new field of research. In order to get good training results therefore evaluation should be performed. For the creation of such systems the questions arising are related to the general process of generation, visualization, evaluation and its psychological background. An important aspect of vision based AR is also the robust tracking and initialization of objects for correct augmentation. In this work we present a concept of an entire processing chain, which allows for efficient and automatic generation of such training systems that can also be used for evaluation. We do this in the context of a Lego training system. While explaining the whole process of application generation and usage, we also present a novel approach for robust marker free initialization of colored partly occluded plates and their tracking using one off the shelf monocular camera.