High-fidelity learning-based motion cueing algorithm by bypassing worst-case scenario-based tuning technique

Qazani, Mohammad Reza Chalak; Asadi, Houshyar; Najdovski, Zoran; Alsanwy, Shehab; Zakarya, Muhammad; Alam, Furqan; Ouakad, Hassen M.; Lim, Chee Peng; Nahavandi, Saeid · 2024 · Cognitive Robotics

DOI: 10.1016/j.cogr.2024.07.001

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the inefficiency of traditional motion cueing algorithms (MCAs) in driving simulators, which are typically tuned for worst-case scenarios. This conventional tuning approach limits the algorithm's effectiveness for medium or slow driving motions, leading to suboptimal workspace utilization and increased motion sickness due to mismatches between visual and vestibular cues. The authors propose a high-fidelity, learning-based MCA that bypasses worst-case tuning by utilizing comprehensive motion signal data across various driving intensities. The methodology involves generating motion signals using the Rigs of Rod (RoR) vehicle simulation environment, covering slow, medium, and fast driving strategies. These signals are processed through three classically tuned washout filters (translational, tilt coordination, and rotational) to establish optimal motion cues for each scenario. The input and output data from these filters are then used to train three separate multilayer perceptron (MLP) neural networks. These MLPs replace the traditional high-pass and low-pass filters within the MCA structure. The models are trained offline using Bayesian regularization and validated using Simulink/MATLAB, incorporating a human vestibular model to assess motion sensation accuracy. Results demonstrate that the proposed learning-based MCA outperforms the classical washout filter tuned for worst-case scenarios. The MLP models achieved high correlation coefficients (CC) and R-square values, indicating strong fidelity in reproducing motion cues. Specifically, the correlation coefficient for sensed specific force (SSF) between the real vehicle driver and the simulator user was 0.7390 for the learning-based model, compared to 0.4901 for the classical filter. Furthermore, the root mean square error (RMSE) for SSF was 22.40% lower in the learning-based MCA. The study also found that the learning-based approach allows for more efficient use of the simulator’s platform workspace, avoiding the over-compensation often seen in classical filters. The significance of this work lies in its ability to enhance simulator realism and reduce motion sickness by adapting to varying driving conditions rather than relying on fixed, worst-case parameters. By replacing fixed filter parameters with adaptive neural networks, the proposed method provides a more efficient and accurate solution for motion cueing. This approach offers a practical improvement for simulator applications in automotive, aviation, and research sectors, where precise motion reproduction is critical for training and testing.

Key finding

The proposed learning-based motion cueing algorithm significantly reduces motion sensation errors and improves workspace efficiency compared to classical washout filters tuned for worst-case scenarios.

Methodology

simulation_modeling

Sample size: 10

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via author_sweep_intake on 2026-05-28.

StageOutcomeToolModelPromptAttemptsCompleted
discover success author_sweep 2 2026-05-28
archive success canonical_url 11 2026-06-06
extract success cached 3 2026-06-10
clean success clean 1 2026-06-04
chunk success chunk 1 2026-06-04
embed success embed Qwen/Qwen3-Embedding-8B 1 2026-06-04
enrich success 1 2026-05-28
promote success 1 2026-06-04
summarize success llm qwen3.6-27b-prismaquant summ-v5 2 2026-06-10
tag success vector_similarity 15 2026-06-11
verify success 2 2026-06-10

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).