Virtual Reality Laboratory
The VR Lab of the A.I. group at Bielefeld University has been founded in 1995. The central research lines of the lab are intelligent computer graphics and human-machine interaction.
The research focus of this laboratory is on intelligent human-machine interaction in virtual worlds involving natural language, gesture, and facial expression. What are the roles played by emotion, attention in space, or knowledge of the dialog partner? How can they be captured in cognitive architectures? Technologies range from realtime processing of speech, gesture and gaze, over acoustic, visual and tactile stimulation to immersive interaction with virtual agents. System prototypes are developed that model cognitive performances and exploit them in technical applications.
Originating from research focuses on natural language processing, expert system techniques, and human machine communication, the AI Lab at Bielefeld found a new mission in 1995 when we started the AI and Computer Graphics Lab, centered around the VIENA and CODY projects which had both started in 1993. Our major motivation was that knowledge-based systems and further AI techniques can be used to establish an intuitive communication link between humans and highly interactive 3D graphics. Realizing that Virtual Reality proves a most comprehensive communication media and a multi-modal interface to a host of multimedia applications, it was a natural step to further extend the Lab's mission to Virtual Reality where we are collaborating with the Institute for Media Communication (IMK), now Fraunhofer IAIS, which is a part of the Fraunhofer Gesellschaft. Our focus is now to build Intelligent VR Interfaces that yield highly interactive construction and design environments by the support of knowledge-based techniques. The SGIM project - which extended our efforts to include multimodal speech and gesture interfaces.
After the success of several projects, the planning phase for a major lab reconstruction and upgrade started back in the year 2000. With the launch of the Virtuelle Werkstatt, the already installed one-sided projection systems wall and responsive workbench are now complemented and substituted by several new systems, e.g., a three-sideed large-screen projection system similar to the CAVE or the cyberstage systems called TRI-SPACE and a portable large screen VR-system named CYKLOOP.
The picture sequence gives an impression of one of our virtual environments, generated and projected using the TRI-SPACE: A user is immersed in a computer generated scene where he is collaborating with our embodied conversational agent MAX. An overview of all necessary "hardware reconstruction work" concerning the lab can be found on the the holodeck construction site page which documents the stages during some of the harware setup and the first tests. Besides the overall research direction that centers around human cognition and strongly AI influenced methods for human-machine interaction using multimodal input and output channels, some of us - which can best be described as VR infected - additionally explore VR specific research topics. This includes rendering techniques, image generation, display systems as well as systems for human movement tracking. Concerning these topics, there are still lots of deficiencies with today's VR related hardware. To name just a few, the commonly used electromagnetic tracking systems are often unprecise, use cables or have quite a bit of latency. The latter is true not only for the tracking, but for the whole image generation process and rendering. To achieve stereo views, analog cathod ray based video projectors with limited brightness and contrast usually drive shutter glasses in active stereo mode. Therefore the display environment must be very dark and the time-sequential stereo allows only quite low frame rates. To overcome these problems in the new lab, we are exploring new and alternative components required for VR. The current equipment includes:
- TRI-SPACE, an immersive VR-display system providing 3 stereoscopic screens using 6 digital JVC D-ILA projectors by 3Dims (see Figure 1)
- Interaction Space, three portrait HD panels by NEC
- CYKLOOP, a portable large screen VR-display with one monoscopic screen.
- HMD, a Head Mounted Display Z800 by eMagin
- 8 channel digital 3D audio system
Holodeck Compute Cluster
- for interactive computer graphics and multimodal interaction
- Infiniband Network
- 7 nodes, 56 CPU cores, 14 GPUs
- Intel Xeon Dual Quad (8 Cores), 2.66 GHz, 8 GB RAM
- NVIDIA Quadro FX 5600
- NVIDIA Quadro FX 5600
- 6 nodes, 24 (+24 HT) CPU cores, 6 GPUs
- Intel Corei7 950, 3.0 GHz, 6 GB RAM
- NVIDIA GTX 580, 3 GB RAM
- 1 node, 4 CPU cores, 1 GPU, 6 channel render server in one node
- Intel Core2 Quad (x4), 2.66 GHz, 8 GB RAM
- 3x NVIDIA GeForce 8800 GTX
- An optical (cable-less) tracking system (see Figure 2) by ART
- Two optical (cable-less) 5 finger gloves by ART
- Two optical (cable-less) 3 finger gloves with tactile feedback (see Figure 3) by ART
- One eye tracker (see Figure 4), ViewPoint PC-60, BS007, from Arrington Research
- Modelling: Blender, 3D Studio MAX
- Visualization & VR:
- Communication: WAS - WBS Agent System
- Sound: FMOD by Firelight Technologies Pty, Ltd.
- Further tools: ECLIPSE, CLIPS, Elk Scheme, Vanilla Sound Server (VSS) of the NCSA
Projectssee projects page.
Ipke Wachsmuth (E-Mail: email@example.com) or
Thies Pfeiffer, 2011-11-30