Analyzing Differences between Questionnaire-Surveyed Driving Style and Revealed Style from Driving Simulation Experiments

Variations in driving styles among individuals lead to distinctive decision-making and behavioral characteristics. Investigating these characteristics of interaction behavior is vital for accurately predicting the behavior of autonomous vehicles. Many researchers employ questionnaires to determine driving styles, yet few validate whether questionnaire outcomes align with data analysis results. In this study, we establish a driving simulation experiment platform to collect driver behavior data in a controlled setting, particularly focusing on unprotected left-turn intersections. We design a questionnaire to assess drivers’ subjective perceptions of their driving styles, aligning group interaction experiments accordingly. We compile vehicle trajectory data (encompassing transverse and longitudinal velocity, acceleration, etc.) and driver operation data (including pedal force and steering wheel angle) from 134 experiments involving 10 drivers. By comparing self-reported driving styles with model outcomes and questionnaire results, our analysis reveals substantial differences in the composition and proportions of driving styles, especially between straight-line driving and left turns. It becomes evident that even drivers identifying as conservative can exhibit occasional aggressiveness.