Measuring Distraction Potential of Operating In-Vehicle Devices

摘要

Three experiments were conducted to explore the feasibility of adapting existing protocols to assess in-vehicle information systems (IVIS) in production vehicles. Two low-fidelity driving simulators were used: the Lane Change Task (LCT) and the STISIM-Drive combined with the Peripheral Detection Task (PDT). The Rating Scale Mental Effort (RSME) workload rating scale and FaceLab eye tracker were also used. Experiment 1 combined simulator driving with laboratory tasks (visual search and short-term memory scanning), which allowed secondary task load to be systematically varied. Metrics sensitive to changes in visual load included LCT Mean Deviation and the following STISIM/PDT metrics: Car-Following Delay, Standard Deviation of Lane Position (SDLP), Steering Entropy, PDT Mean Response Time, and Proportion of Correct PDT responses. Among objective metrics, only PDT Mean Response Time was sensitive to changes in cognitive load associated with the (auditory/vocal) memory-scanning task. Experiment 2 used real-world secondary tasks performed with a factory-installed navigation system. Secondary tasks differed by input modality (manual vs. voice) and task complexity (destination entry vs. selecting previous destinations). STISIM/PDT metrics, including SDLP, Steering Entropy, and Proportion of Correct PDT Responses were sensitive to task differences, as was the LCT Mean Deviation. STISIM/PDT metrics were more sensitive than LCT metrics to differences in both experiments. The RSME subjective rating scale was sensitive to most differences, while eye position data was not sufficiently reliable to allow computation of eye glance-based metrics. Experiment 3 used an established test-track protocol to determine whether measures obtained in a real driving situation exhibited greater sensitivity to potential distraction effects than those obtained in the simulation laboratory. Several laboratory simulator measures were more sensitive to secondary task load differences than corresponding test track measures. We concluded that the STISIM/PDT test venue offers sufficient sensitivity for development of a portable test of IVIS distraction potential in production vehicles for visual/manual tasks; improved sensitivity is needed for assess effects of cognitive distraction. We identified technical problems and questions about test validity to be addressed in subsequent developmental work.