Making Augmented Reality Practical on Mobile Phones , Part 1

Daniel Wagner and Dieter Schmalstieg, Graz University of Technology In the past few years, mobile phones have become an increasingly attractive platform for augmented reality (AR; see Figure 1 for some examples). According to Gartner, 1.22 billion mobile phones were sold in 2008. Some forecasts estimate that this number will rise to 1.8 billion units in 2012, of which 800 million are expected to be smartphones. Although not all these devices are open for custom software development, the trend toward open software environments for mobile phones seems inevitable. In 2003, we started an AR framework for mobile phones (for a glimpse of other research on AR for mobile phones, see the sidebar). We intended its fi rst generation as primarily a proof of feasibility. The second generation was an attempt to port a fully featured PC-based AR framework, Studierstube 4 (http://studierstube.org), to a phone platform. You can port existing applications and make them run on mobile phones. However, as we had to painfully experience ourselves, this approach typically produces slow, bloated, and unstable software. Optimally using phones’ scarce resources requires different algorithms and architectural decisions than for PCs, leading to a complete reengineering of an existing solution. So, for the third generation, Studierstube ES, we largely abandoned compatibility requirements and added new elements to the design, such as an asymmetric client-server technique, that are specifi c to mobile devices.3 In this fi rst installment of our two-part tale of Studierstube ES and what we’ve learned along the way, we describe the mobile phone platform’s restrictions and how our software architecture allows fast development of mobile phone AR applications.