Driving simulator validation studies: A systematic review

Driving simulators (DS) serve as pivotal platforms for the rigorous testing of transportation systems and vehicles, offering a safe, controllable experimental environment with features like design visualization, scenario virtualization, and test data quantification. The validation of simulator experiments relies on the realism of the driving experience and scenario fidelity, crucial for assessing data reliability and result credibility. With the advent of autonomous driving technologies, the frequency of DS utilization has seen a marked expansion. Nonetheless, the discourse surrounding DS validation remains nascent, lacking a consolidated framework of standards and evaluative methodologies. This review endeavors to synthesize existing scholarly discourse and reports on the validation of driving simulators, further probing into the suitability of various driving scenarios and tasks. Common scenarios include car-following, lane-changing, and acceleration/deceleration, while tasks encompass human-machine co-piloting, takeover scenarios, and emergency evasion, considering driver conditions such as fatigue and distraction. Extracting universal indicators from various scenarios, including longitudinal and lateral velocities, accelerations, and trajectories, the paper summarizes the experimental workflow and commonly used statistical testing methods and psychophysiological monitoring devices for driving simulator validation. Considering the multidimensional factors influencing validation, this study discusses the relationships between simulation fidelity, degrees of freedom (DOF), and simulator sickness, proposing reference standards for driving simulator validation. This effort aims to advance the establishment of evaluation norms for simulation-based transportation and vehicle research, ensuring scientific rigor and empirical validity.