The Fully Reconfigurable CAVE
Leaving the CAVE and entering the FRAVE
Product developers, vehicle design engineers and trainee pilots increasingly utilize 3D worlds, operating in virtual space created by realistic images in real time. For this purpose they use virtual reality systems. Product designers harness time-consuming procedures in prototype construction. Only then are they able to assess the results of their work in a comprehensive manner. In a three-dimensional model world, they are able to do so instantly and can experience how the product fits into its natural surroundings. Design alterations can be visualized immediately, saving time and cutting the costs associated with the development process. A popular system to achieve the above is the so-called CAVE (Cave Automatic Virtual Environment). This consists of between three and six projection surfaces that create a walk-in space. Video projectors are used to visualize the calculations and applications in real time and in 3D. The nearly-closed space allows for intense immersion in virtual reality.
The FRAVE can do more. Not only does it offer a certain degree immersion but it can also be used in a variety of ways thanks to its flexible, modular structure. An engineer wants to enter the 3D world to be able to envisage the interior design of a vehicle. A researcher wants to visualize his or her measurement or simulation data, while a manager uses it as a presentation space, as Dr. Marcus Tönnis, scientist at the TU München Faculty of Informatics explains.
The FRAVE is made up of ten plasma screens with a screen diagonal of 65 inch which can be arranged in different ways. When they form a floor and an enclosure on three sides, the user is immersed in a virtual exploratory world. The screens at the sides can be opened wide, with a tracking system on the screens automatically adapting the image display to the movement of the side sections. The side sections can even be disconnected from the system entirely.
Standard hardware makes virtual reality systems more affordable As the FRAVE consists of of-the-shelf components, it is significantly less expensive than the CAVE, an advantage that could promote more widespread use of Virtual Reality Systems. Another important benefit of the FRAVE is its a smaller footprint. Since the CAVE normally uses back projection, a lot of room is needed behind the projection surfaces. It requires at least 8 x 8 x 8 meters, while a space of 3 x 3 x 3 meters is sufficient for the FRAVE, thereby facilitating installation and relocation.
What happens when I change this parameter?
Visualization environments gain an increasing importance, not only for virtual reality applications but also for visualiza- tion of simulation data. While fully immersive visualization systems are important or almost mandatory for presence in virtual environments, simulation systems do not necessarily require full immersivity. A simulation researcher might need a wide field of view, but eventually wants to inspect a specific effect at a certain location and then might want to sit down for further examination and discussion. Situations like See that effect over there? What happens when I change this parameter? are frequent occurrences when exploring simulated spaces.
Researchers at the TUM Faculty of Informatics use the FRAVE to view simulation data and their dependence on parameters.. For example, the landscape of Saudi Arabia is displayed virtually as part of the Virtual Arabia project being run in collaboration with the King Abdullah University of Science and Technology (KAUST). Unlike existing virtual globes, like Google Earth, this system is able to show images above and below the earth's surface. As part of the C02 Sequestration research project, the FRAVE is being used to simulate CO2 separation and storage processes in order to optimize crude oil extraction.
6x Building block:
- 2x (resp. 1x) Panasonic TX-P65VT20E, 65" Full HD, 120 fps (60 HZ for 3D)
- 2x Intel Xeon E5630, (4 Core, 12M cache, 2,53 GHz)
- 2x 12 GB EEC DDR3-1066 RAM
- 1x Nvidia Tesla C2070 (15 SP, 6 GB RAM)
- 1x Nvidia Quadroplex 7000 (2x16 SP, 2x6 GB RAM)
- 1x Mellanox Connect-X2 VPI HCA (40 Gbps IB or 10 Gbps ET)
3x Post-processing nodes :
- IBM x3550 M3 Server Blade
- 2x Intel Xeon X5690, (6 Core, 12M cache, 3,46 GHz)
- 2x 48 GB ECC DDR3-1333 RAM
- 1x Mellanox Connect-X2 VPI HCA (40 Gbps IB or 10 Gbps ET)
- 2x Switch Mellanox IS5030 for 1410 36-Port Managed QDR IB
- 2.88 Tbps switching capacity
- 36 ports with 40 Gbps per port
- 9x HCA Card: Mellanox Connect-X2 VPI HCA
- 3D viewing via Active Shutter Technology
- rigid body tracking with 8x OptiTrack S250e infrared cameras
- Phantom Premium 6 DOF Haptic Device
- 5DT Data Glove Ultra
- Microsoft Kinect and mobile devices
Current Software Development on the FRAVE
|How to load terabytes of very detailed height and image data so fast that a smooth fly over a large are is possible? The terrain renderer developed at TUM's computer graphics chair does exactly that in a divide and conquer manner. It splits the data in pieces and loads only high detailed blocks for areas close to the viewer and for the rest lower resolution data is used.||
|Simulations nowadays depend on several parameters of which many combinations are of interest when understanding a new simulation setup. The problem is the long simulation time needed by a single simulation. The computational steering approach running on the FRAVE uses the power of the graphic accelerators to deliver simulation results in real time to the screens.||
|A single building consists of thousands of construction pieces. What about a block of buildings or an entire city? Can we store this amount of data and easily search through it? The infrastructure explorer deals with issue by using smart data structures to organize and deliver accurate infrastructure information to architects and city simulations.||
FRAVE Opening Hours
The FRAVE is located here and can be visited during opening hours (Monday 10-12 AM and Thursday 10-12 AM) or upon request.
FRAVE In The News
FRAVE In Publications
- D. Butnaru, G. Buse and D. Pflüger, A Parallel and Distributed Surrogate Model Implementation for Computational Steering, In Proceeding of the 11th International Symposium on Parallel and Distributed Computing (ISPDC), IEEE, 2012.