Learning in the freespace driving simulator: An exploratory study with active safety systems
Abstract
Driving simulators have been used in transportation research for decades, but human subject learning is mostly unknown. Most researchers understand the need for a familiarization phase, but the point at which learning is complete and “normal” driver behavior begins is not entirely understood. This study attempts to address this issue while also assessing driver trade-off behavior in response to active safety system implementation as it relates to risk compensation theories. In other words, possible changes in driver behavior, specifically speed and lane keeping, will be evaluated around the addition of active safety systems. Four subjects drove a 27-mile four-lane road in the Freespace driving simulator repeatedly over a five week timeframe. Data related to lane keeping, speed, and encroachment incidents was collected and analyzed subject-by-subject, by scenario, and using statistical modeling. Subject 1 proved to be an aggressive driver who took many risks, but actually did not crash. Subject 2 appeared to be a slower learner, never taking a large number of risks but consistently improving. Subject 3 began the study with two user-caused crashes and afterwards appeared to be attempting to find the right level of confidence to match the skill level. Subject 4 began the experiment having already used the simulator multiple times during the calibration and testing phase. This experience resulted in interesting results where this subject rarely tested the limits of the simulator and ended up with the least number of total encroachment incidents. Although each subject learned differently, there were definitive personal trends in lane keeping, speed, and encroachment incidents. Long-term trends were also different between subjects, essentially blending the line between learning and potential natural fluctuations in attention, risk perception, confidence, and driving. The data fluctuation may be explained by risk allostasis theory, where drivers adapt to an ever-changing perceived task difficulty through feelings of risk. Also, there were no obvious effects of active safety systems on driver behavior and performance as applied in this study. Future research involving more subjects and simulation runs would undoubtedly give a better picture of learning in a simulator and the potential effect of active safety systems.
Degree
M.S.C.E.
Advisors
Tarko, Purdue University.
Subject Area
Behavioral psychology|Civil engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.