Investigation of hybrid tracking basics in the context of Augmented Reality

In Augmented Reality (AR) real images and virtual images are overlaid in real time to produce the illusion of an enhanced world perception. AR needs a very exact determination of the camera position, in order to be able to combine the real and artificial objects correctly. To achieve the necessary accuracy optical tracking methods have been developed, which determine the camera position from markers with sub-pixel accuracy directly from the pictures. Those methods are however unstable with strong movements or if few marker are visible in the picture. Other tracking systems, like magnetic and acoustic trackers could help to solve those problems, but they do not supply the necessary accuracy. Hybrid tracking techniques consist of using different tracking systems at the same time which allows to compensate the weaknesses of each one. Therefore, a hybrid tracking system can satisfy the high tracking requirements of AR. In this work, the coupling of different tracking systems is examined. An emphasis is thereby on the combination of optical and acoustic trackers and optical and compass-based systems. In the first step the theory of a hybrid tracking method is developed and the requirements and constraints are described. Afterwards, this hybrid tracking method is implemented using a vision based tracking algorithm and a inertial-acoustic tracker. In order to test the different configurations of the method before apply them to the real application, a simulation environment has been created using Matlab. The development takes place in C/C++ under UNIX (SGI-IRIX) and Matlab. Finally the results obtained are presented and the conclusions and future possible enhancements discussed. The method can perform the tracking of the camera in a fast and robust way.