Defining and Assessing Endpoints in Driving Simulator Tasks

• Activity Number: 317
• Type: Topic Contributed
• Date/Time: Tuesday, August 9, 2005 : 2:00 PM to 3:50 PM
• Sponsor: General Methodology
• Author(s): Qian Shi*+ and Laura Stierman and Jeffrey Dawson and Matthew Rizzo
• Companies: The University of Iowa and Northwestern University and The University of Iowa and The University of Iowa
• Keywords: Crash plots ; Data reduction ; Validation ; Driving performance assessment

Abstract

Driving simulators are capable of generating driving scenarios that replicate key aspects of real-life situations. In these scenarios, driving performance and safety errors can be assessed under controlled and replicable conditions that are challenging yet safe to the driver and investigator. For example, a virtual vehicle can be triggered to pull out unexpectedly in front of a subject's vehicle to assess whether the subject responds aptly and quickly enough to avoid a crash. The outcome can be defined in terms of the appropriateness of the maneuver or as a time-to-event variable (such as time until first reaction). Using the digitized data captured in a simulator, these outcomes can be defined objectively and measured. However, subjective review of the sample individual responses is generally required to verify the usefulness of the chosen outcome definitions and the correctness of the data reduction procedures. We illustrate this process for specific tasks and for measures of overall driver safety using data from a fixed-base simulator known as SIREN.