5. Conclusion

While the ability to increase visibility in a virtual environment without significantly impairing depth estimation is an exciting outcome of these experiments, the successful application of software methods to hardware limitations validates our earlier assertion about the usefulness of in-code solutions. That is, the flexibility of software design permits VE system designers to account for the technological limitations of the hardware while minimizing the effect on human perception and task performance. This approach has not been widely used yet has proved to be a most effective solution.

Furthermore, the attempt to preserve realism via adherence to human characteristics has also been shown to be advantageous. Using human perceptual and performance data to design systems has long been the domain of human factors engineers, but the implementation of VE systems clearly needs extensive investigation in this area. The issue of realism in VE simulators is a nontrivial one, and an attempt should be made to better define what constitutes a good, immersive simulation.

Unfortunately, to answer the questions about realism and adherence to the real-world requires experimentation with transfer tasks. Transfer task experimentation would demonstrate how skills learned in a VE carry over to a real world environment. The nature of the problem in the OOD simulator prevented any such kind of experimentation (moving a buoy around in a bay and asking a person to estimate the distance is unreasonably difficult). If we were able to test both the VE and the real world performance on a task, we could gauge the effectiveness of the VE for simulating that task. We could also assess whether making the VE more "realistic" had an effect on performance in both the VE and the real world. We could, in this way, define what makes a VE realistic, and how we can go about attaining realism.

Furthermore, the implications of training in a VE could be examined more fully with a task that could be performed in both the synthetic world and the real world. We could discover the ways in which a simulator needs to be realistic in order to train a given task. The best training methodology and the most realistic (in terms of human perception) may be totally different. While this seems unlikely, certainly the manner in which a VE simulator needs to be realistic should be examined both as a function of performance in the VE and as a function of transfer to the real world.

The issue of determining realism aside, using human performance as a guideline to design and implement VE systems is clearly important. VEs are the most human-centered systems ever developed by modern engineers and have incredible potential for a wide variety of applications. This potential can only be realized by continuing research into the particular eccentricities of virtual environment systems and their interaction with human beings.

On to next section: References