Home // International Journal On Advances in Systems and Measurements, volume 3, numbers 3 and 4, 2010 // View article

Assessment of Simulator Fidelity and Validity in Simulator and On-the-road Studies

Authors:
Andreas Riener

Keywords: Simulator validity/fidelity, Real-driving studies, Trace-driven simulation, Driver-vehicle interaction (DVI), Feedback modalities, Performance evaluation

Abstract:
A lot of research groups all over the world have tried to relate results from driving simulator studies to real driving behavior. A solution, e. g. in form of a conversion table, would be of great value. Unfortunately, status quo is that even with expensive, high fidelity simulators the validity of results cannot be guaranteed. One reason for this is that a person's behavior cannot be described by mathematical rules and depends, beside the task of interaction, on several subsidiary influence factors. Starting with an elaborate review of driving validity and fidelity constraints, the aim of this paper is to summarize on our research responding to the question to what extent driving simulators can be used to serve as cheap and easy realizable environments for simulating on-the-road behavior. The purpose of the studies was to determine (i) whether or not it is in general possible to approximate real driving with simulator studies, (ii) situation and modality dependent correction or scale factors to deduce real reaction times from simulation, and (iii) further requirements, parameters, and restrictions to be satisfied for succeeding high fidelity studies. Two user studies were conducted, a low fidelity trace-driven simulation in a lab environment and a on the road driving experiment. Recorded reaction times were compared in order to assess the validity of data generated in these experimental series. The events were, in the case of simulation, triggered trace-driven or, in the real driving experiment, manually activated by the experimenter and notifications were forwarded to the driver using a random assignment of one of the modalities vision, hearing, or touch. Results indicate that drivers responds faster to steering requests in the driving simulator compared to real driving. The explanation for this difference can most likely be derived from the fact that test persons were less demanded in the first (artificial) compared to the second (real) setting. When analyzing data on individual notification channel basis, it can be observed that (i) the order of channels with respect to average response times is the same in both settings (vibro-tactile, visual, auditory) and (ii) the reaction time differences are almost uniformly distributed. Prospective work in comparative studies is projected to happen in two directions, on improvements in the behavior of the low fidelity simulator to become as close to reality as possible and in the utilization of high fidelity driving simulators to directly relate real driving results to.

Pages: 110 to 124

Copyright: Copyright (c) to authors, 2010. Used with permission.

Publication date: April 6, 2011

Published in: journal

ISSN: 1942-261x