Exploring the role of knowledge graphs and large language models in autonomous vehicle safety
Tsinghua University Press
image:
The framework and structure of this review
view moreCredit: Communications in Transportation Research
In a new comprehensive review, researchers from Tsinghua University, Xiaomi EV, Beijing Institute of Technology, and University of Electronic Science and Technology of China examine how these two AI paradigms are being applied across the development, validation, and operation of autonomous driving systems, discuss their strengths and limitations, and further review their emerging synergy for more robust, interpretable, and trustworthy autonomous driving systems.
The team published their study in Communications in Transportation Research (https://doi.org/10.26599/COMMTR.2026.9640023).
Two complementary AI paradigms for autonomous driving safety
Autonomous vehicles must operate in highly dynamic and uncertain traffic environments. Although recent advances in sensing and learning have improved perception and control, current systems still face major challenges in scene understanding, risk reasoning, rare scenario handling, and explainability.
In the study, the research team examined two important AI approaches with complementary strengths. Knowledge graphs represent structured knowledge such as traffic rules, object relationships, causal chains, and expert driving experience, enabling explicit reasoning and traceable decision support. Large language models, in contrast, are more flexible in semantic understanding, contextual generalization, and reasoning over open-ended situations.
Comparison and synergy of these two AI paradigms for autonomous driving safety
The review shows that knowledge graphs and large language models offer different advantages for autonomous driving safety. Knowledge graphs are well suited for organizing structured knowledge and supporting explicit, interpretable reasoning, especially in tasks involving traffic rules, causal relations, and expert knowledge. Large language models, in contrast, are more flexible in semantic understanding, contextual reasoning, and handling open-ended situations, which gives them strong potential in complex and uncertain traffic environments.
Rather than treating them as competing approaches, the study further highlights their growing synergy. Researchers are increasingly exploring how structured knowledge can improve the transparency and reliability of large language models, while the flexible reasoning ability of large language models can expand how knowledge graphs are queried, understood, and applied. This emerging combination points to a promising direction for building more robust, interpretable, and trustworthy autonomous driving systems.
DOI Link:
https://doi.org/10.26599/COMMTR.2026.9640023
About Communications in Transportation Research
Communications in Transportation Research was launched in 2021, with academic support provided by Tsinghua University and China Intelligent Transportation Systems Association. The Editors-in-Chief are Professor Xiaobo Qu, a member of the Academia Europaea from Tsinghua University, and Professor Xiaopeng (Shaw) Li from University of Wisconsin–Madison. The journal mainly publishes high-quality, original research and review articles that are of significant importance to emerging transportation systems, aiming to serve as an international platform for showcasing and exchanging innovative achievements in transportation and related fields, fostering academic exchange and development between China and the global community.
It has been indexed in SCIE, SSCI, Ei Compendex, Scopus, CSTPCD, CSCD, OAJ, DOAJ, TRID and other databases. It was selected as Q1 Top Journal in the Engineering and Technology category of the Chinese Academy of Sciences (CAS) Journal Ranking List. In 2022, it was selected as a High-Starting-Point new journal project of the “China Science and Technology Journal Excellence Action Plan”. In 2024, it was selected as the Support the Development Project of “High-Level International Scientific and Technological Journals”. The same year, it was also chosen as an English Journal Tier Project of the “China Science and Technology Journal Excellence Action Plan Phase Ⅱ”. In 2024, it received the first impact factor (2023 IF) of 12.5, ranking Top1 (1/58, Q1) among all journals in "TRANSPORTATION" category. In 2026, its 2025 IF was announced as 12.7, maintaining the Top1 position (1/66, Q1) in the same category.
From Volume 6 (2026), Communications in Transportation Research will be published by Tsinghua University Press on the SciOpen platform with the official journal website at https://www.sciopen.com/journal/2097-5023. We kindly request that all new manuscript submissions be made through the journal’s submission system at https://mc03.manuscriptcentral.com/commtr. For any submission-related inquiries, please contact the Editorial Office at commtr_e@mail.tsinghua.edu.cn.
Journal
Communications in Transportation Research
Article Title
Enhancing autonomous vehicle safety with knowledge graphs and large language models: Comprehensive review
One-pedal driving in electric vehicles: A boon for automation, but a challenge for manual control
Tsinghua University Press
image:
The graph illustrates the experimental speed profile used to evaluate one-pedal driving (OPD) and two-pedal driving (TPD). It features four distinct segments of speed fluctuations and transitional deceleration phases, simulating both high-speed and low-speed traffic conditions to test how following vehicles adapt to varying levels of disturbance.
view moreCredit: Communications in Transportation Research
With the rapid global shift toward electric vehicles (EVs), one-pedal driving (OPD) has become a standard feature. In this mode, the accelerator pedal alone manages both acceleration and deceleration. While celebrated for improving energy efficiency by up to 22% in urban environments, its specific impact on traffic flow stability and car-following behavior remained largely unexplored until now.
To address this gap, a research team at the University of Georgia conducted field experiments using a fleet consisting of one lead internal combustion engine vehicle and three following EVs equipped with OPD and Adaptive Cruise Control (ACC).
The team published their study in Communications in Transportation Research (https://doi.org/10.26599/COMMTR.2026.9640030).
“Our findings highlight a fundamental difference in how regenerative braking interacts with the control source,” says Handong Yao, an Assistant Professor in the College of Engineering at the University of Georgia. “In manual driving, the high sensitivity of the system can lead to overcorrections by drivers. However, when managed by automated systems like ACC, that same sensitivity allows for incredibly precise and stable longitudinal control”.
Manual Control: The Learning Curve
Under manual control, the study found that OPD produces more pronounced speed oscillations and sharper decelerations compared to traditional two-pedal driving (TPD). This is particularly evident among drivers who are novices to the one-pedal paradigm.
- Increased Variability: Drivers struggled with fine speed modulation at low speeds, resulting in wider distributions of spacing and acceleration.
- Quicker Response: Deceleration under OPD is initiated more rapidly because there is no mechanical lag from switching between the accelerator and brake pedals.
- Adaptation Needs: The researchers suggest that during the early stages of OPD adoption, drivers may require additional training or supportive safety features, such as visual feedback, to mitigate operational noise and reduce potential risks.
ACC Integration: Unlocking Stability
The results took a dramatic turn when automation was introduced. When integrated with ACC, one-pedal driving yielded narrower distributions of speed, acceleration, and spacing than traditional two-pedal driving.
- Smoother Flow: The ACC system leverages the zero-transition time of regenerative braking to respond more conservatively and smoothly to speed changes by precisely regulating regenerative torque.
- Platoon Stability: This seamless control effectively attenuates speed perturbations throughout the vehicle string, suggesting that OPD-equipped automated vehicles could significantly improve overall traffic flow efficiency and stability.
“The synergy between one-pedal driving and smart infrastructure or automated systems is where the real potential lies,” explains Tianle Zhu, a Ph.D. student and the study’s first author. “By aligning these natural regenerative curves with automated guidance, we can move toward a more stable, energy-efficient, and safer mobility ecosystem”.
DOI Link:
https://doi.org/10.26599/COMMTR.2026.9640030
About Communications in Transportation Research
Communications in Transportation Research was launched in 2021, with academic support provided by Tsinghua University and China Intelligent Transportation Systems Association. The Editors-in-Chief are Professor Xiaobo Qu, a member of the Academia Europaea from Tsinghua University, and Professor Xiaopeng (Shaw) Li from University of Wisconsin–Madison. The journal mainly publishes high-quality, original research and review articles that are of significant importance to emerging transportation systems, aiming to serve as an international platform for showcasing and exchanging innovative achievements in transportation and related fields, fostering academic exchange and development between China and the global community.
It has been indexed in SCIE, SSCI, Ei Compendex, Scopus, CSTPCD, CSCD, OAJ, DOAJ, TRID and other databases. It was selected as Q1 Top Journal in the Engineering and Technology category of the Chinese Academy of Sciences (CAS) Journal Ranking List. In 2022, it was selected as a High-Starting-Point new journal project of the “China Science and Technology Journal Excellence Action Plan”. In 2024, it was selected as the Support the Development Project of “High-Level International Scientific and Technological Journals”. The same year, it was also chosen as an English Journal Tier Project of the “China Science and Technology Journal Excellence Action Plan Phase Ⅱ”. In 2024, it received the first impact factor (2023 IF) of 12.5, ranking Top1 (1/58, Q1) among all journals in "TRANSPORTATION" category. In 2026, its 2025 IF was announced as 12.7, maintaining the Top1 position (1/66, Q1) in the same category.
From Volume 6 (2026), Communications in Transportation Research will be published by Tsinghua University Press on the SciOpen platform with the official journal website at https://www.sciopen.com/journal/2097-5023. We kindly request that all new manuscript submissions be made through the journal’s submission system at https://mc03.manuscriptcentral.com/commtr. For any submission-related inquiries, please contact the Editorial Office at commtr_e@mail.tsinghua.edu.cn.
Journal
Communications in Transportation Research
Article Title
Investigating the car-following behavior of electric vehicles with one-pedal driving
No comments:
Post a Comment