EUVIMEDEUVIMED entdecken
Crossref

Adaptive Takeover Time Budget Allocation in Automated Driving: A Physiology-Aware Semantic Agent Framework

Yijun Lu, Yuyin Ma, Jianjin Zhao, Jiong Dong, Jialong Li, Nobukazu Yoshioka, Naoyasu Ubayashi, Hironori Washizaki

IEEE Transactions on Cognitive Communications and Networking · 2026 · 12 · 9038-9051

DOI: 10.1109/tccn.2026.3705802 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Cognitive Communications and Networking, 2026. Verfasst von Yijun Lu und 7 weiteren Autor:innen. Die Arbeit nennt 49 Literaturangaben. Gefördert von: JST BOOST, Japan, collaborative research program between Kioxia Corporation and Waseda University, JSPS KAKENHI.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In conditionally automated driving (SAE Level 3), safe and comfortable control handover requires an appropriate takeover time budget (TOTB) that accounts for both driving context and driver readiness. Existing fixed-budget heuristics have limited adaptability to changing environments and heterogeneous physiological states, while data-driven approaches often provide limited interpretability under noisy or ambiguous physiological cues. To address these limitations, we propose a Physiology-Aware Semantic Agent (P-SA) for context-adaptive TOTB planning. P-SA combines physiological and contextual signals, maps biosignals into normalized cognitive tokens through subject-specific calibration, and integrates these tokens with Quality of Takeover (QoT) and Quality of Experience (QoE) objectives via multimodal prompting to a Large Language Model. The inferred policy is further projected into deterministic and physically feasible actions under safety constraints. We evaluate P-SA on a high-fidelity cyber–physical digital-twin platform with human-in-the-loop takeover tasks. Results show that P-SA provides a favorable balance among safety, efficiency, and comfort. It achieves zero collision events across all P-SA takeover tasks ( $0/132$ ), improves kinematic stability by 32% relative to the aggressive fixed-budget baseline, reduces redundant takeover time by approximately 4.0 s compared with the conservative baseline, and lowers peak physiological stress. These findings suggest that physiology-aware semantic reasoning is a promising direction for adaptive and human-aware takeover time allocation in HITL vehicular systems.

Förderung

  • JST BOOST, Japan
  • collaborative research program between Kioxia Corporation and Waseda University
  • JSPS KAKENHI
  • General Program of the Natural Science Foundation of Xinjiang Uygur Autonomous Region
  • Basic Research Funds of Universities in the Autonomous Region
  • Natural Science Foundation of Henan Province
  • Key Scientific Research Project Plan of Henan Provincial Colleges and Universities
  • Henan Province International Science and Technology Cooperation Project–Cultivation Project

Zitieren

APA

Lu, Y., Ma, Y., Zhao, J., Dong, J., Li, J., Yoshioka, N., Ubayashi, N., & Washizaki, H. (2026). Adaptive Takeover Time Budget Allocation in Automated Driving: A Physiology-Aware Semantic Agent Framework. IEEE Transactions on Cognitive Communications and Networking, 12, 9038-9051. https://doi.org/10.1109/tccn.2026.3705802

Vancouver

Lu Y, Ma Y, Zhao J, Dong J, Li J, Yoshioka N, et al. Adaptive Takeover Time Budget Allocation in Automated Driving: A Physiology-Aware Semantic Agent Framework. IEEE Transactions on Cognitive Communications and Networking. 2026;12:9038-9051. doi:10.1109/tccn.2026.3705802

BibTeX
@article{Lu2026,
  author = {Lu, Yijun and Ma, Yuyin and Zhao, Jianjin and Dong, Jiong and Li, Jialong and Yoshioka, Nobukazu and Ubayashi, Naoyasu and Washizaki, Hironori},
  title = {Adaptive Takeover Time Budget Allocation in Automated Driving: A Physiology-Aware Semantic Agent Framework},
  journal = {IEEE Transactions on Cognitive Communications and Networking},
  year = {2026},
  volume = {12},
  pages = {9038--9051},
  doi = {10.1109/tccn.2026.3705802}
}
Kostenlos starten →