Nt1330 Unit 3 Assignment 1

However, these two solutions are just extreme examples of how work may be allocated among a server and handheld client. Depending on circumstances, solutions in between these extremes may be useful and necessary. If one limits the discussion to a typical AR system which uses a single video source for both tracking and video see-through display, the processing pipeline is composed of the following main tasks: video acquisition, tracking, application computation, rendering, display. Offloading some of these tasks to a computing server is an instance of horizontally distributed simulation, and it is established knowledge that a scalable solution (many clients, many servers etc.) requires cautious use of the available network bandwidth. Communication

Both projects again use the method describe in Figure 4(d). Shibata's work aims at load balancing between client and server - the weaker the client, the more tasks are outsourced to a server. It can therefore vary between all situations described in Figure 4. ULTRA uses PDA-based AR to support maintenance workers, but concentrates on augmenting "snapshot" still images. In the absence of realtime tracking for infrastructure independence it performs all tasks natively (Figure 4a). In 2003 ARToolKit to the PocketPC was developed, the first fully self-contained PDA AR application. This platform was used in a peer to peer game. Möhring et al. were the first to successfully target a consumer smartphone for mobile AR. The scarce processing power of the target platform allowed only a very coarse estimation of the object's pose on the screen. Henrysson ported ARToolKit to the Symbian platform and created the first twoplayer AR game on current-generation smart phones. Summarizing these developments one can conclude that there is no ideal solution for systems with scarce processing

The user’s or device’s pose must be measured accurately, robustly and in real-time. Most mobile phones today are equipped with built-in cameras, which naturally lend to use computer vision based approaches for tracking. Tracking fiducial markers is a common strategy to achieve robustness and computational efficiency simultaneously. The ARToolKitPlus library was developed based on the well known open source ARToolKit library. It was ported to the Windows CE environment, optimized for the mobile phone platform and heavily extended with features that support mobility, such as automatic thresholding or large amounts of markers. The very latest version of ARToolKitPlus (also known as Studierstube Tracker) is a complete redevelopment that is no longer related to ARToolKit on a source code basis and improves several outstanding issues such as