Here is an extract of our thesis proposal we wrote for the application of a DAAD-scorlarship

The idea of Hybrid Tracking is to combine different tracking technologies to make a better assumption about the current context of the environment. Another advantage is to have a redundant tracking-system. If one tracking component fails another component now delievers the context-data.

This is our first scenario: A user using GPS enters a house. GPS does not work inside building, so another tracker has to deliver spacial information. This could be a camera attached to the user, which detects an ARToolkit marker at the entrance of the house. Now the system knows that it has to switch to another tracking-component. The challenge of this is, that the user is not aware of the tracking method that is used at the moment.

This piece of work is done together with Michael Siggelkow (MichaelSiggelkow^?). Although this is not a group diploma thesis we will try to concentrate in different aspects of Hybrid Tracking Applications and technologies. My focus will be to keep an eye on the usability of complex application using hybrid tracking to generate input for the software system.

Our idea was to layer such applications into three components:

• top-layer: Application Layer
• middle-layer: Integration Layer
• bottom-layer: Tracking-Layer

The Tracking Layer is responsible to deliver raw tracking data to the Integration Layer. It may consist of many different trackers or one method is used for tracking different things (one camera tracks the user, another tracks a certain object). This data is commited to the Integrations Layer which tries to combine the data to one context-representation for the Application Layer. The Integration Layer is also responsible to deliver context-information that is requested by the application. The Application Layer on top uses the requested context-representation to perform its task.

Michael will focus on the Tracking Layer and I will focus on the Application Layer. We will then try to integrate our components in the Integration Layer.

We already did some brainstorming about possible aspects of our thesis:

• Develop interfaces for the layers (e.g develop a context representation language, middle- to top-layer)
• Evaluation of different tracking technologies and combinations of them
• Quality of service for tracking applications (e.g. the Application layer requests tracking data with a certain quality depending on the task to perform. The Integration Layer then tries to aggregate tracking data which delivers that quality.)
• Evaluation of the usability (this will be my focus)
• Find possible hybrid-tracking applications
• Implementation into the DWARF-framework (Augmented Reality Framework)

Michael had a great idea^? for an application scenario: The idea is to develop a 3D-modelling system. The user is able to draw geometries with a stick in the air. The stick and the user might be tracked by different trackers. The drawing is automatically imported in a 3D modelling tool (like Maya or 3D Max). I think optical tracking like this was used in the making of "Lord of the Rings". The motion of the Gollum creature was created by a real person who was tracked by cameras and then the information was rendered in a 3D modelling tool.

But our ideas are only suggestions. It only gives a direction for our thesis. We still have to look how we can realize parts of our proposal at the Hitlab. Or we will even find more interesting questions in New Zealand during our research.

-- FelixLoew - 24 Sep 2004