Perception and action in virtual environments: Human performance in virtual driving simulators.

摘要

The real issue and challenge in VR today is not the display, but how to reconcile a person's expectations of reality. Virtual environment (VE) systems are becoming commonplace in research settings. In many cases, virtual environments are used to simulate existing real environments. A common example of this is the use of vehicle simulators (e.g., driving, and flight simulators) for training and assessment of performance in the corresponding real vehicles (cars and aircraft, respectively). In training and assessment, the virtual environment has little inherent interest. Rather, the virtual environment is valuable to the extent that performance in it resembles performance in the simulated real-world environment. This differs from virtual environment applications that do not simulate extant, real worlds. Examples include numerous entertainment applications involving interplanetary travel, fantasy worlds, and so on. Three VE studies with a group of commonalities and unique differences are contrasted here in an attempt to address and assess the components which can be held as reliable indicators of RW performance. I will also examine the limitations of interpreting performance data gathered from VE simulations.; Simulation Validation addressed primarily lane position and gap affordances in comparison to a matched RW environment. Forward Looking Blindspots addressed the perception of threat in a VE and A-pillar issues in automotive design. Deer Avoidance also addressed perception of threat and driver response to enhanced warning signage. The primary foci of these experiments were the validation or more precisely the fidelity of the VE participant's performance relative to the real world (RW) which may to some degree be driven by habitual or at least not-quite-conscious expectations. VEs are limited in many aspects. Wrap around (non head mounted) simulators cannot currently produce glare, parallax, changing resolution in response to accommodation, and any number of RW visual effects. They cannot legally be made to induce pain or threat of injury. Why then should performance data still be used to reflect on or compare to RW performance?