Towards objective motion cueing tuning in driving simulators for vehicle dynamics evaluation

QC 2026-08-12

Abstract

Vehicle manufacturers face increasing pressure to reduce development time and costs while maintaining vehicle performance. Although computer-aided engineering has enabled a largely virtual development process, a key limitation remains: the integration of human subjective assessment. Ride and handling characteristics are inherently perceptual and still rely on physical testing, thus delaying feedback to later development stages.

Moving-base driving simulators offer a way to introduce the human into the virtual loop and enable early subjective evaluation of vehicle models. However, their effectiveness depends on the motion cueing,which maps vehicle motion to the limited workspace of the simulator. These algorithms are still largely tuned subjectively, which requires physical reference data and offers limited assurance of fidelity. This restricts their usefulness in early-phase development, particularly during winter testing when conditions are difficult and testing opportunities are limited.

The goal of this work is to establish an objective approach to motion cueing evaluation and development based on physically interpretable models combining vehicle dynamics, simulator motion and human perception. This leads to a shift in perspective, wherein motion cueing is treated as a vehicle dynamics representation problem rather than a simulator control problem.

The thesis demonstrates that objective evaluation is possible using simple linear models, which are applied to analyse cueing fidelity and to determine the positioning of the longitudinal axis of rotation, showing good agreement with both optimisation-based and subjective methods. The results show that the motion reference point has a first-order influence on perceived motion and can introduce systematic distortions.

Similarly, tilt coordination is shown to improve immersion while altering perceived vehicle characteristics. Moreover, conventional vestibular models are extended with head-neck dynamics and gaze stabilisation to better capture perception. This highlights that motion cueing is constrained by visualinertial coupling and that cabin-fixed vestibular models alone are insufficient.

The work further shows that vehicle motion should be partitioned into path-related and vehicle-relative path components. Path-related motion is unbounded and must be limited by simulator constraints, while vehicle-relative path motion is bounded and can be reproduced more accurately. This enables selective filtering that preserves key motion components, in contrast with conventional approaches wherein all states are treated uniformly. Using this framework, separating slip angle feedback from high-pass filtered cues improves yaw representation, and separating vehicle roll and pitch from road-induced motion enables more consistent cueing compared to classical and model predictive approaches. It is also shown that motion scaling inherently distorts vehicle dynamics and that meaningful scaling needs to preserve key state relationships, thus leading to constraints on how lateral acceleration and yaw rate should be scaled.

Finally, simulator limitations are addressed explicitly by using reachability-based methods that enable objective offline tuning of motion cueing within the available workspace to maximise the simulator usage for a given motion cueing and manoeuvre.

Overall, the thesis establishes a perception-aware and vehicle-centric framework for motion cueing. By exploring how cueing alters perceived vehicle characteristics and providing objective methods for evaluation and design, the work supports the use of driving simulators for reliable vehicle dynamics assessment in early-phase development.

Link to DiVA