An NMPC-Based False-Cue-Suppressing Motion Cueing Algorithm for 9-DoF Motion Simulators

Motion cueing algorithms reproduce the perceived motion of vehicles within the limited workspace of driving and flight simulators. Nine-DoF simulators enlarge this workspace considerably yet expose three difficulties: rectangular task space envelopes leave much of the reachable set unused, sustained accelerations produce vestibular false cues, and the nonlinear per-stage linearization of model predictive control is the real-time bottleneck. To address these difficulties, this paper proposes separate strategies within a nonlinear model predictive cueing algorithm for such platforms. Constraints are imposed directly on the leg lengths obtained from the kinematic inverse, replacing the rectangular envelope and making the full reachable workspace available. A deadband penalty, referenced to vestibular perception thresholds and the current reference magnitude, suppresses cues that contradict the reference in sign. Finally, the Newton–Euler dynamics are approximated by an extended dynamic mode decomposition model with a neural network residual, which turns per-stage linearization into a constant-matrix read, shortening the solving time to meet the real-time control requirement at an acceptable accuracy cost. Closed-loop simulations on four flight maneuver classes show sustained false cues driven back below the perception threshold while the reachable workspace is fully exploited.