The aim of this project is to explore a collaborative prototype for the integration of GIS sensor data into a distributed architecture for Virtual Reality.

The aim of this project developed within the scope of the research proposal of Martin Witzel is to explore a collaborative prototype for the integration of GIS sensor data into a distributed architecture for Virtual Reality. Sensor data of various origin and classified domains are provided using a topic subscription approach.

The system's main features are:

• Distribution of sensor stations.
• Graphical administration of distributed sensor stations.
• Collaborative visualization of sensor data.
• Continuous integration of sensor data.
• Integration of terrain into an immersive environment.
• User friendly, easy-to-use desktop interface.
• 3D contextual visualization of WFS data.

The system provides support for collaborative GIS sessions based on the architecture depicted in the following picture:



The several existing components are inter-connected based on a distributed Mixed Reality/Augmented Reality Design Review architecture which includes a flexible XML message-based communication layer named OSGA (Open Source Groupware Architecture) to share messages and data between software components and applications.

Terrain Support

To achieve a strong support of higher level terrain visualization, we chose to integrate the Open Source Virtual Terrain Project (VTP) functionalities into our system by porting it to the Open Source Scenegraph Library OpenSG (link to port). Doing so, the clients take advantage of rendering techniques like various continuous level of detail implementations and paging of elevation and imagery.

Manipulation of sensor data

A graphical editor, which allows creation/import/export/manipulation of sensor data, servers as a real-time administrative interface to control the provision of topics/sensorial data and, at the same time, it is used as an interface to common GIS applications.

The standard VTP graphical editor, used for viewing and processing geospatial data (VTBuilder, 2D), has been extended through the development of the so-called SensorBuilder. This tool allows the definition and graphical manipulation of sensor data, boundaries and attributes. Following the architecture of the standard VTP graphical editor we use a layer-based architecture in which data are classified by the user into a specific contexts. Layers are thus assigned a context like water, structures, elevation and imagery.

A client application, responsible for the rendering, decides according to the context which data (timevarying, static) is relevant for the actual examination. Additionally, terrain cultures like forests, buildings and streets along with their geometrical and semantical attributes are provided by using interactive queries to a Web Feature Service/PostGIS database and translated to a 3D representation.
These queries are broadcasted to all clients in the framework to assure at all times synchronicity of the examination data.

Administration of sensor data

The concept of administrating sensor stations via the Sensor-
Builder relay application requires us to support two distinct approaches: handling of raw, unsupervised sensor data, and additionally managed, manipulated sensor data through insertion of a real-time administration component into the communication pipeline.

After the user logs in to the communication backbone, SensorBuilder listens on the SENSOR channel for the presence of sensor stations and sensor domains and keeps track of changes by storing available data sets for each station in a dictionary. In case the user links an observation value, e.g. humility, to a sensor boundary, SensorBuilder acts as a managing component for this particular station and the observation chosen value domain. The pair {station, value domain} is then marked as managed.

]

This way, we allow both direct input from a sensor source and additionally supervised, modified input via an administration component.

Session management is enabled by storing the messages in a relational database to achieve persistency and recalling of the analysis process.