Prediction-Loss-Guided Long-Tail Augmentation and Uncertainty-Aware Reinforcement Learning for Car-Following Control
IEEE Internet of Things Journal · 2026-08 · 13 (16) · 36930-36946
DOI: 10.1109/jiot.2026.3700902 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Internet of Things Journal, 2026-08. Verfasst von Hongrui Cao und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 31 Literaturangaben. Gefördert von: State Key Laboratory of High-Performance Precision Manufacturing, Fundamental Research Funds for the Central Universities, State Key Laboratory of Automotive Chassis and Bionics Open Fund Integration.
Abstract (Originalsprache)
Reinforcement learning (RL) has demonstrated great potential for autonomous car-following control by directly optimizing key performance metrics, including safety, efficiency, and ride comfort. This study proposes an integrated RL-based car-following framework that combines trajectory prediction, reshaping of long-tail distributions, and uncertainty-aware decision-making. The trajectory prediction loss is redefined as a quantitative metric for scenario rarity and risk, facilitating a distribution-level data augmentation strategy that enriches safety–critical scenarios while preserving the naturalness of driving behavior. A probabilistic Gaussian long short-term memory (LSTM) predictor is further employed to model future vehicle motion uncertainty, which is explicitly incorporated into multistep action evaluation through variance-weighted virtual rewards. Experiments on the highD dataset demonstrate that the proposed approach significantly improves robustness and safety, achieving substantial reductions in collision rates compared with baseline RL methods. Cross-dataset evaluation on NGSIM further confirms its generalization capability.
Förderung
- State Key Laboratory of High-Performance Precision Manufacturing
- Fundamental Research Funds for the Central Universities
- State Key Laboratory of Automotive Chassis and Bionics Open Fund Integration
Zitieren
APA
Cao, H., Cai, L., Yue, M., Bao, H., & Guan, H. (2026). Prediction-Loss-Guided Long-Tail Augmentation and Uncertainty-Aware Reinforcement Learning for Car-Following Control. IEEE Internet of Things Journal, 13(16), 36930-36946. https://doi.org/10.1109/jiot.2026.3700902
Vancouver
Cao H, Cai L, Yue M, Bao H, Guan H. Prediction-Loss-Guided Long-Tail Augmentation and Uncertainty-Aware Reinforcement Learning for Car-Following Control. IEEE Internet of Things Journal. 2026;13(16):36930-36946. doi:10.1109/jiot.2026.3700902
BibTeX
@article{Cao2026,
author = {Cao, Hongrui and Cai, Lei and Yue, Ming and Bao, Haiyun and Guan, Hsin},
title = {Prediction-Loss-Guided Long-Tail Augmentation and Uncertainty-Aware Reinforcement Learning for Car-Following Control},
journal = {IEEE Internet of Things Journal},
year = {2026},
volume = {13},
number = {16},
pages = {36930--36946},
doi = {10.1109/jiot.2026.3700902}
}
EUVIMED