Optimizing simulator performance specifications using driver performance measurements

The quantity of driving simulator applications around the world appears to be increasing dramatically. Most of these applications study driver behaviors. However, very few studies evaluate how different simulator performance properties might affect driver behaviors. So, the potential exists for differences in simulator performances to produce conflicting conclusions about driver behaviors. This paper documents a process that correlates simulator performance specifications to driver training requirements through analyzing data collected while performing real training. This analysis applies no model of human performance. Instead it applies measurements of actual driver performance collected while training more than 1000 law enforcement drivers in emergency vehicle operations (EVO) over a two-year period. Because the data is both objective and application-specific, it can be used to correlate driver performance measurements with simulator performance and thereby identify critical specifications for the simulation. This paper illustrates a process of analysis that may be used to correlate driving simulator specifications for different types of simulator applications, provided that the application collects objective driver performance measurements. Ironically, the measurement data needed to develop simulator procurement specifications is only produced by the application itself! This analysis takes a step toward resolving that conundrum.