Chapter 10: Surrogate Methods and Measures
Handbook of Driving Simulation for Engineering, Medicine, and Psychology
Surrogate Methods and Measures
Linda S. Angell, Touchstone Evaluations, Inc.
The Problem. Beginning in about 2000, a need emerged for methods with which to assess the effects of advanced information systems on driver workload early in product development—surrogate methods that were valid, practical, and applicable before on-road driving studies could be done with instrumented vehicles. Surrogate methods that emerged included techniques such as visual occlusion, peripheral detection tasks, the lane-change test, and other techniques. As these methods have been studied, a variety of questions emerged about their possible use and role in the context of driving simulation. Role of Driving Simulators. Driving simulators are an important tool for examining responses of drivers to the driving scenarios they present. Simulators are themselves a kind of surrogate—a surrogate for the overall driving experience on the road, in scenarios—of interest. In that context, this chapter explores what role other surrogate methods might play—how they are different, as well as how they can be used together with simulation—in the study of driving behavior. Key Results of Studies for Appropriately Applying Surrogates and Simulations. Studies in different venues (lab, simulator, track, and road) have been done which provide key findings that are salient for how best to use surrogates and simulations together in research. These findings are reviewed, with an emphasis on building robustness into research designs. Scenarios and Dependent Variables. Examples of the sorts of surrogates that can be incorporated in simulation are discussed, along with surrogates that are implemented as simulations. Issues are identified, along with opportunities for innovation through simulation. Generality. Both surrogates and simulations can be powerful tools in their own right, but the key to unlocking their power lies in the clarity of the conceptualization that underlies the research question driving their application. This is a general and fundamental point, but a very critical one to advancing the start of the art.
Surrogate Methods, Metrics, Product Development, Product Testing, Redline
• There are different types of surrogate methods, and careful distinctions between types must be made when designing experiments.
• Some seek to emulate key cues of the driving environment, and are used to create a “driving-like” experience for a driver. (These would usually be considered simulations.)
• Others seek instead to evoke a set of processes in the human operator that are similar to those elicited during the actual driving behavior/s of interest—as a means of making these processes observable and measurable. (These are surrogate methods which do not qualify as simulations.)
• Surrogate methods may be embedded within simulations, and used to advantage.
• Simulations may themselves be used as surrogate methods.
• The terms “method” and “metric” are different, and it is useful to distinguish between them. Surrogate “methods” refer to the procedures, equipment, and protocols through which data are acquired. Surrogate “metrics” refer to specific measures of performance or behavior that are generated by a method (e.g., time to complete a task, total glance time to perform a task). In addition, a “metric” may be associated with decision criteria, or “redlines,” that represent values on that metric above or below which performance is deemed “unacceptable.” Such “redlines” are often used in setting requirements for product design.
• Having a clear conceptualization with which to guide research using simulations or surrogates is the key to harnessing the power of these techniques. A thorough understanding of driver perceptual, cognitive, and motoric processes—together with key research questions—will benefit the choice and implementation of method (surrogate or simulation) and metric.
Angell, L., Auflick, J., Austria, P. A., Kocchar, D., Tijerina, L., Biever, W., . . . Kiger, S. (2006). Driver workload metrics task 2 final report and appendices (Report: DOT HS 810 635). Washington, DC: National Highway Traffic Safety Administration, US Department of Transportation.
Engström, J. (2008, October). The peripheral detection task and related methods: Theoretical and methodological issues. Paper presented at the Driver Metrics Workshop. San Antonio, TX.
Victor, T. W., Engström, J., and Harbluk, J. L. (2008). Distraction assessment methods based on visual behavior and event detection. In M. A. Regan, J. D. Lee, and K. Young (Eds.), Driver distraction: Theory, effects, and mitigation (pp. 135–168). London: Taylor and Francis.