Motion cueing in driving simulators for research applications

This research investigated the perception of self-motion in driving simulation, focussing on the dynamic cues produced by a motion platform. The study was undertaken in three stages, evaluating various motion cueing techniques based on both subjective ratings of realism and objective measures of driver performance. Using a Just Noticeable Difference methodology, Stage 1 determined the maximum perceptible motion scaling for platform movement in both translation and tilt. Motion cues scaled by 90% or more could not be perceptibly differentiated from unscaled motion. This result was used in Stage 2‟s examination of the most appropriate point in space at which the platform translations and rotations should be centred (Motion Reference Point, MRP). Participants undertook two tracking tasks requiring both longitudinal (braking) and lateral (steering) vehicle control. Whilst drivers appeared unable to perceive a change in MRP from head level to a point 1.1m lower, the higher position (closer to the vestibular organs) did result in marginally smoother braking, corresponding to the given requirements of the longitudinal driving task. Stage 3 explored the perceptual trade-off between the specific force error and tilt rate error generated by the platform. Three independent experimental factors were manipulated: motion scale-factor, platform tilt rate and additional platform displacement afforded by a XY-table. For the longitudinal task, slow tilt that remained sub-threshold was perceived as the most realistic, especially when supplemented by the extra surge of the XY-table. However, braking task performance was superior when a more rapid tilt was experienced. For the lateral task, perceived realism was enhanced when motion cues were scaled by 50%, particularly with added XY-sway. This preference was also supported by improvements in task accuracy. Participants ratings were unmoved by changing tilt rate, although rapid tilt did result in more precise lane control. Several interactions were also observed, most notably between platform tilt rate and XY-table availability. When the XY-table was operational, driving task performance varied little between sub-threshold and more rapid tilt. However, while the XY-table was inactive, both driving tasks were better achieved in conditions of high tilt rate. An interpretation of these results suggests that without the benefit of significant extra translational capability, priority should be given to the minimisation of specific force error through motion cues presented at a perceptibly high tilt rate. However, XY-table availability affords the simulator engineer the luxury of attaining a slower tilt that provides both accurate driving task performance and accomplishes maximum perceived realism.