Abstract
We have developed a modular virtual environment platform for movement research and rehabilitation. The system uses several networked computers running Linux to share computation. An electromagnetic tracker is the primary position tracker and both a head-mounted display and stereo goggles are used for visual display. System software is written in a combination of C++, JAVA, and Python and makes considerable use of the open-source toolkits VR Juggler and OpenSceneGraph. These are integrated with additional toolkits and custom modules written specifically for the study of motor control and rehabilitation. The system performs well with low latency, accurate calibration, and a consistently high graphics update rate. Preliminary applications have confirmed that the system is a powerful tool for sensory-motor investigation and has considerable potential as a tool for neurorehabilitation. Its primary advantage over other systems is its ability to utilize different display and input devices, and run a range of experiments simply by changing XML configuration files. Additionally, the use of powerful open-source libraries provides a feature-rich foundation for advanced features and low-cost duplication. Further work and experiments are needed to extend, further validate, and fully utilize this platform.