The behavioural validation of driving simulators as research tools: a case study based on the Leeds Driving Simulator

The objectives of this thesis was to provide researchers with a scientitically-based guide for
interpreting driver behaviour results obtained on a fixed-base driving simulator and to
provide guidance on how the Leeds Advanced Driving Simulator (LADS) could be
modified to overcome any deficiencies that were detected. However. objectives of any
simulator validation study are directly related to the specitic driving task under
investigation. our ability to perform a similar task in the field (for the comparison of the
results between the two environments) and the existing configuration capabilities of the
simulator.
To achieve the objectives of this study, driver behaviour was investigated at the control
level under different road geometry and oncoming traffic conditions using the LADS. Speed
and lateral displacement in terms of mean and standard deviation were chosen to represent
driver behaviour. They were measured under free-flowing conditions on a rural A road. The
objectives of the study were fulfilled by comparing observational uncontrolled real road
data with experimental simulator data and by evaluating the differences between the two
environments using the absolute and relative validity criteria. It was found that LADS is
relatively valid in terms of speed and lateral position. It was also found that higher speeds
are developed in the simulator where speed in not confined by the road geometry and
simulator subjects drive significantly closer to the edge of the road compared to their real
road counterparts irrespective of the road geometry and the oncoming traffic conditions.
The face validity of the simulator was examined using subjective data obtained from
questionnaires relative to the realism and ease of controlling the simulator. Subjects
commented that the least realistic features of the simulator were the braking and steering
systems. Subjects were classiffied to "good" and "poor" according to their responses
regarding the simulator face validity. It was found that "good" subjects behave slightly
better compared to "poor" subjects when driving the simulator.