2023年9月24日から28日までスペインのビルバオで開催される第26回 IEEE International Conference on Intelligent Transportation Systems (ITSC 2023)での発表のため、研究室から提出した5つの研究論文が受理されました。
IEEE Intelligent Transportation Systems Society主催のITSCは、世界中の研究者、業界専門家が、一堂に会する毎年恒例の主要イベントです。この会議は、高度道路交通システム(ITS)領域における理論、解析的・数値的シミュレーションやモデリング、実験、先進的な展開、ケーススタディなどの新展開を発表する場として機能しています。
採択された各論文は、インテリジェント交通システムのユニークな側面を掘り下げています。ビジョントランスフォーマーによるパーソナライズされた危険なシーンの理解から、衝突回避のためのオクルージョンを考慮した経路計画まで、幅広いトピックをカバーしています。さらに、我々の論文のうち2本は、オープンソースプロジェクトです。ToST(東京SUMO交通シナリオ)とFlowsim(都市規模のコネクテッド自律走行車向けモジュール式シミュレーションプラットフォーム)は、論文発表と同時にソースコードとデータをオープン化します。
ITSC 2023で我々の研究を発表する機会に感謝するとともに、我々の革新的な研究をITSコミュニティと共有したいと考えています。今年の9月にビルバオで開催されるITS国際会議にぜひご参加ください。
Paper
Bao, Naren, Carballo, Alexander, Tsukada, Manabu, Takeda, Kazuya, "Personalized Causal Factor Generalization for Subjective Risky Scene Understanding with Vision Transformer", In: The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023), Bilbao, Bizkaia, Spain, 2023.Proceedings Article | Abstract | Links | BibTeX
@inproceedings{Bao2023,
title = {Personalized Causal Factor Generalization for Subjective Risky Scene Understanding with Vision Transformer},
author = {Naren Bao and Alexander Carballo and Manabu Tsukada and Kazuya Takeda},
doi = {10.1109/ITSC57777.2023.10422148},
year = {2023},
date = {2023-09-24},
urldate = {2023-09-24},
booktitle = {The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023)},
address = {Bilbao, Bizkaia, Spain},
abstract = {This paper presents a framework to understanding subjective driving scene perception by Vision Transformer for Environmental Feature Extraction within a Causal Modeling Analysis method. By leveraging vision transformer models, informative features are extracted from video camera images capturing the surrounding environment. Through the causal analysis, the causal effects of these variables on subjective risk perception are explored, shedding light on the factors influencing individuals' perception of driving risk. The findings demonstrate understanding of environmental features and individual difference on risk perception, providing a deeper understanding of risky scene perception. The paper concludes with this approach unifies selective attentional phenomena can improve the scene understanding for subjective perception in real-world driving scenarios aiming to enhance driving safety based on the identified causal factors. },
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper presents a framework to understanding subjective driving scene perception by Vision Transformer for Environmental Feature Extraction within a Causal Modeling Analysis method. By leveraging vision transformer models, informative features are extracted from video camera images capturing the surrounding environment. Through the causal analysis, the causal effects of these variables on subjective risk perception are explored, shedding light on the factors influencing individuals' perception of driving risk. The findings demonstrate understanding of environmental features and individual difference on risk perception, providing a deeper understanding of risky scene perception. The paper concludes with this approach unifies selective attentional phenomena can improve the scene understanding for subjective perception in real-world driving scenarios aiming to enhance driving safety based on the identified causal factors.
Yamazaki, Yuji, Tamura, Yasumasa, Defago, Xavier, Javanmardi, Ehsan, Tsukada, Manabu, "ToST: Tokyo SUMO traffic scenario ", In: The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023), Bilbao, Bizkaia, Spain, 2023.Proceedings Article | Abstract | Links | BibTeX
@inproceedings{Yamazaki2023,
title = {ToST: Tokyo SUMO traffic scenario },
author = {Yuji Yamazaki and Yasumasa Tamura and Xavier Defago and Ehsan Javanmardi and Manabu Tsukada},
url = {https://github.com/dfg-lab/ToSTScenario},
doi = {10.1109/ITSC57777.2023.10422517},
year = {2023},
date = {2023-09-24},
urldate = {2023-09-24},
booktitle = {The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023)},
address = {Bilbao, Bizkaia, Spain},
abstract = {In recent years, research, development, and demonstrations aimed at the societal implementation of autonomous driving have attracted increasing attention. Localization, which involves obtaining information of the surrounding environment from sensor data and estimating the vehicle's position, is necessary for realizing autonomous driving. Localization is commonly performed with 3D LiDAR as a sensor owing to its high measurement accuracy and immunity to ambient light conditions, which allow for precise localization. However, when the surrounding area has distinctive features, localization accuracy may decrease. In this study, we proposed a method based on deep learning to predict the localization accuracy for autonomous driving. The overall localization accuracy can be improved by predicting the accuracy of localization using other sensors, such as GNSS and IMU, or pavement markings in areas with poor accuracy. We created a dataset for predicting the localization accuracy using an open-source autonomous driving simulator. In an experiment, we applied the proposed method to the created dataset. Thresholds were set for errors in the x-direction, y-direction, and distance for localization. Predictions with high accuracy and F-values were obtained. The results indicate that the proposed method can accurately predict the localization accuracy. },
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
In recent years, research, development, and demonstrations aimed at the societal implementation of autonomous driving have attracted increasing attention. Localization, which involves obtaining information of the surrounding environment from sensor data and estimating the vehicle's position, is necessary for realizing autonomous driving. Localization is commonly performed with 3D LiDAR as a sensor owing to its high measurement accuracy and immunity to ambient light conditions, which allow for precise localization. However, when the surrounding area has distinctive features, localization accuracy may decrease. In this study, we proposed a method based on deep learning to predict the localization accuracy for autonomous driving. The overall localization accuracy can be improved by predicting the accuracy of localization using other sensors, such as GNSS and IMU, or pavement markings in areas with poor accuracy. We created a dataset for predicting the localization accuracy using an open-source autonomous driving simulator. In an experiment, we applied the proposed method to the created dataset. Thresholds were set for errors in the x-direction, y-direction, and distance for localization. Predictions with high accuracy and F-values were obtained. The results indicate that the proposed method can accurately predict the localization accuracy.
Lin, Pengfei, Javanmardi, Ehsan, Nakazato, Jin, Tsukada, Manabu, "Occlusion-Aware Path Planning for Collision Avoidance: Leveraging Potential Field Method with Responsibility-Sensitive Safety", In: The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023), 2023.Proceedings Article | Abstract | Links | BibTeX
@inproceedings{Lin2023b,
title = {Occlusion-Aware Path Planning for Collision Avoidance: Leveraging Potential Field Method with Responsibility-Sensitive Safety},
author = {Pengfei Lin and Ehsan Javanmardi and Jin Nakazato and Manabu Tsukada},
url = {https://arxiv.org/abs/2306.06981},
doi = {10.1109/ITSC57777.2023.10422621},
year = {2023},
date = {2023-09-24},
urldate = {2023-09-24},
booktitle = {The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023)},
series = {Bilbao, Bizkaia, Spain},
abstract = {Collision avoidance (CA) has always been the foremost task for autonomous vehicles (AVs) under safety criteria. And path planning is directly responsible for generating a safe path to accomplish CA while satisfying other commands. Due to the real-time computation and simple structure, the potential field (PF) has emerged as one of the mainstream path-planning algorithms. However, the current PF is primarily simulated in ideal CA scenarios, assuming complete obstacle information while disregarding occlusion issues where obstacles can be partially or entirely hidden from the AV's sensors. During the occlusion period, the occluded obstacles do not possess a PF. Once the occlusion is over, these obstacles can generate an instantaneous virtual force that impacts the ego vehicle. Therefore, we propose an occlusion-aware path planning (OAPP) with the responsibility-sensitive safety (RSS)-based PF to tackle the occlusion problem for non-connected AVs. We first categorize the detected and occluded obstacles, and then we proceed to the RSS violation check. Finally, we can generate different virtual forces from the PF for occluded and non-occluded obstacles. We compare the proposed OAPP method with other PF-based path planning methods via MATLAB/Simulink. The simulation results indicate that the proposed method can eliminate instantaneous lateral oscillation or sway and produce a smoother path than conventional PF methods.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Collision avoidance (CA) has always been the foremost task for autonomous vehicles (AVs) under safety criteria. And path planning is directly responsible for generating a safe path to accomplish CA while satisfying other commands. Due to the real-time computation and simple structure, the potential field (PF) has emerged as one of the mainstream path-planning algorithms. However, the current PF is primarily simulated in ideal CA scenarios, assuming complete obstacle information while disregarding occlusion issues where obstacles can be partially or entirely hidden from the AV's sensors. During the occlusion period, the occluded obstacles do not possess a PF. Once the occlusion is over, these obstacles can generate an instantaneous virtual force that impacts the ego vehicle. Therefore, we propose an occlusion-aware path planning (OAPP) with the responsibility-sensitive safety (RSS)-based PF to tackle the occlusion problem for non-connected AVs. We first categorize the detected and occluded obstacles, and then we proceed to the RSS violation check. Finally, we can generate different virtual forces from the PF for occluded and non-occluded obstacles. We compare the proposed OAPP method with other PF-based path planning methods via MATLAB/Simulink. The simulation results indicate that the proposed method can eliminate instantaneous lateral oscillation or sway and produce a smoother path than conventional PF methods.
Tao, Ye, Javanmardi, Ehsan, Nakazato, Jin, Tsukada, Manabu, Esaki, Hiroshi, "Flowsim: A Modular Simulation Platform for Microscopic Behavior Analysis of City-Scale Connected Autonomous Vehicles", In: The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023), Bilbao, Bizkaia, Spain, 2023.Proceedings Article | Abstract | Links | BibTeX
@inproceedings{Tao2023c,
title = {Flowsim: A Modular Simulation Platform for Microscopic Behavior Analysis of City-Scale Connected Autonomous Vehicles},
author = {Ye Tao and Ehsan Javanmardi and Jin Nakazato and Manabu Tsukada and Hiroshi Esaki},
url = {https://arxiv.org/abs/2306.05738},
doi = {10.1109/ITSC57777.2023.10421900},
year = {2023},
date = {2023-09-24},
urldate = {2023-09-24},
booktitle = {The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023)},
address = {Bilbao, Bizkaia, Spain},
abstract = {As connected autonomous vehicles (CAVs) become increasingly prevalent, there is a growing need for simulation platforms that can accurately evaluate CAV behavior in large-scale environments. In this paper, we propose Flowsim, a novel simulator specifically designed to meet these requirements. Flowsim offers a modular and extensible architecture that enables the analysis of CAV behaviors in large-scale scenarios. It provides researchers with a customizable platform for studying CAV interactions, evaluating communication and networking protocols, assessing cybersecurity vulnerabilities, optimizing traffic management strategies, and developing and evaluating policies for CAV deployment. Flowsim is implemented in pure Python in approximately 1,500 lines of code, making it highly readable, understandable, and easily modifiable. We verified the functionality and performance of Flowsim via a series of experiments based on realistic traffic scenarios. The results show the effectiveness of Flowsim in providing a flexible and powerful simulation environment for evaluating CAV behavior and data flow. Flowsim is a valuable tool for researchers, policymakers, and industry professionals who are involved in the development, evaluation, and deployment of CAVs. The code of Flowsim is publicly available on GitHub under the MIT license. },
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
As connected autonomous vehicles (CAVs) become increasingly prevalent, there is a growing need for simulation platforms that can accurately evaluate CAV behavior in large-scale environments. In this paper, we propose Flowsim, a novel simulator specifically designed to meet these requirements. Flowsim offers a modular and extensible architecture that enables the analysis of CAV behaviors in large-scale scenarios. It provides researchers with a customizable platform for studying CAV interactions, evaluating communication and networking protocols, assessing cybersecurity vulnerabilities, optimizing traffic management strategies, and developing and evaluating policies for CAV deployment. Flowsim is implemented in pure Python in approximately 1,500 lines of code, making it highly readable, understandable, and easily modifiable. We verified the functionality and performance of Flowsim via a series of experiments based on realistic traffic scenarios. The results show the effectiveness of Flowsim in providing a flexible and powerful simulation environment for evaluating CAV behavior and data flow. Flowsim is a valuable tool for researchers, policymakers, and industry professionals who are involved in the development, evaluation, and deployment of CAVs. The code of Flowsim is publicly available on GitHub under the MIT license.
Yosodipuro, Nicholaus Danispadmanaba, Javanmardi, Ehsan, Nakazato, Jin, Tamura, Yasumasa, Defago, Xavier, Tsukada, Manabu, "Mixed-traffic Intersection Management using Traffic-load-responsive Reservation and V2X-enabled Speed Coordination", In: The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023), Bilbao, Bizkaia, Spain, 2023.Proceedings Article | Abstract | Links | BibTeX
@inproceedings{Yosodipuro2023,
title = {Mixed-traffic Intersection Management using Traffic-load-responsive Reservation and V2X-enabled Speed Coordination},
author = {Nicholaus Danispadmanaba Yosodipuro and Ehsan Javanmardi and Jin Nakazato and Yasumasa Tamura and Xavier Defago and Manabu Tsukada},
url = {https://www.researchgate.net/profile/Jin-Nakazato/publication/374470825_Mixed-traffic_Intersection_Management_using_Traffic-load-responsive_Reservation_and_V2X-enabled_Speed_Coordination/links/651ee8d63ab6cb4ec6bde79a/Mixed-traffic-Intersection-Management-using-Traffic-load-responsive-Reservation-and-V2X-enabled-Speed-Coordination.pdf},
doi = {10.1109/ITSC57777.2023.10422248},
year = {2023},
date = {2023-09-24},
urldate = {2023-09-24},
booktitle = {The 26th edition of the IEEE International Conference on Intelligent Transportation Systems (ITSC 2023)},
address = {Bilbao, Bizkaia, Spain},
abstract = {Vehicle-to-everything (V2X) communication enables connected autonomous vehicles (CAVs) to share information and generate optimal decisions. The networking abilities of CAVs have led to the development of unsignalized autonomous intersection management (AIM) methods that leverage CAVs to significantly improve traffic flows. However, AIM methods assume 100% CAV market penetration, which is currently unrealistic owing to the gradual adoption of CAVs. Therefore, CAVs must share road usage with nonconnected vehicles (NCVs). Thus, we propose a mixed-traffic intersection management method that considers NCVs while ensuring high traffic flow, called traffic-load-responsive reservation for intersection management (TLRRIM). In TLRRIM, the roadside unit (RSU) first classifies vehicles and groups them into clusters before selecting a reservation cluster to cross an intersection. The reservation cluster selection considers both traffic load and crossing urgency. In addition, the RSU utilizes V2X-enabled speed coordination (VESC) for CAVs within the reservation cluster to further improve traffic flow, while utilizing traffic lights to guide NCVs. Simulation-based experiments using OpenCDA and CARLA showed that TLRRIM can increase throughput and reduce waiting time by up to 89.63% and 60.71%, respectively, compared with the fixed-time signaling method. Moreover, adding VESC can increase throughput by 12.21% and reduce waiting time by 10.80%, further enhancing traffic flow. },
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Vehicle-to-everything (V2X) communication enables connected autonomous vehicles (CAVs) to share information and generate optimal decisions. The networking abilities of CAVs have led to the development of unsignalized autonomous intersection management (AIM) methods that leverage CAVs to significantly improve traffic flows. However, AIM methods assume 100% CAV market penetration, which is currently unrealistic owing to the gradual adoption of CAVs. Therefore, CAVs must share road usage with nonconnected vehicles (NCVs). Thus, we propose a mixed-traffic intersection management method that considers NCVs while ensuring high traffic flow, called traffic-load-responsive reservation for intersection management (TLRRIM). In TLRRIM, the roadside unit (RSU) first classifies vehicles and groups them into clusters before selecting a reservation cluster to cross an intersection. The reservation cluster selection considers both traffic load and crossing urgency. In addition, the RSU utilizes V2X-enabled speed coordination (VESC) for CAVs within the reservation cluster to further improve traffic flow, while utilizing traffic lights to guide NCVs. Simulation-based experiments using OpenCDA and CARLA showed that TLRRIM can increase throughput and reduce waiting time by up to 89.63% and 60.71%, respectively, compared with the fixed-time signaling method. Moreover, adding VESC can increase throughput by 12.21% and reduce waiting time by 10.80%, further enhancing traffic flow.