Systematization of Traffic Disturbance Scenarios with Explicit Positional Relationships Among Traffic Participants on General Roads

Daisuke Miura･Hisataka Yoshida･Akiyoshi Hayashi (NTT DATA Automobiligence Research Center)･Mayu Mori (MEITEC)･Yoshitake Kozai (NTT DATA Automobiligence Research Center)･Ryo Hasegawa･Yuki Konishi･Hisashi Imanaga (JARI)

This study defines Pre-Logical Scenarios as an intermediate layer between Functional and Logical scenarios for traffic-disturbance situations on general roads. Targeting interactions with vehicles, motorcycles, bicycles, and pedestrians, we propose a unified six-element representation comprising road geometry, ego-vehicle behavior, ego-vehicle position, surrounding participant behavior, position, and direction. This common structure enables consistent scenario generation across all traffic participant types, facilitating a systematic transition toward Logical scenarios.

Development of Foundational Technologies for Comprehensive Safety Evaluation Against Traffic Disturbances Encountered by Automated Vehicles
-Integrated Extraction of Traffic Laws and Traffic Disturbance Scenarios Corresponding to Road Structures and Signs-

Hisataka Yoshida･Daisuke Miura･Akiyoshi Hayashi (NTT DATA Automobiligence Research Center)･Mayu Mori (MEITEC)･Yoshitake Kozai (NTT DATA Automobiligence Research Center)･Ryo Hasegawa･Yuki Konishi･Hisashi Imanaga (JARI)

This study proposes a method for comprehensively generating Pre-Logical Scenarios for traffic disturbances on general roads and systematically associating real-world constraints - including road structure, physical structures such as guardrails, and traffic-law regulations - with each scenario to enable structured extraction of scenarios relevant to safety evaluation. An automated tool implementing this method was developed as a foundational technology for comprehensive safety evaluation of automated driving systems. Case studies were conducted to demonstrate its practical applicability.

Development of a Real-Time Intersection Risk Prediction Method Utilizing Two-Dimensional Occupancy Grid Maps

Hironori Chaya･Heishiro Toyoda (Toyota Mortor)

This study proposes a real-time risk prediction method for intersections. Using traffic data observed on-site, the proposed method considers signal states and speeds separately for vehicles and pedestrians and efficiently extracts future near-miss risks and their potential hazards based on past similar situations. Such a framework enables us to holistically and simultaneously predict potential conflicts among multiple traffic participants.

A Study of Careful Driving Behavior in Right-Turn Maneuvers at Intersections by General Human Drivers

Yuki Manabe･Toru Kojima･Akihiro Abe･Mamoru Takushima･Kouichi Kitada (NALTEC)

To investigate referential human driving behavior for automated driving vehicles, including harmony with surrounding traffic, a driving simulator study was conducted on cautious driving behavior of general human drivers in situations where they may approach an oncoming vehicle while making a right turn at an intersection.

Study on High-Precision Estimation Methods for Vehicle Trajectory Using Drive Recorder Video and 3D-LiDAR Data

Hideki Matsumura (NALTEC)･Kazuki Tainaka (Tokyo Metropolitan University)･Kotaro Ishizuka (Institute for Traffic Accident Research and Data Analysis)･Motoki Sugiyama (nstitute for Traffic Accident Research and Data Analysis)･Naoyuki Kubota (Tokyo Metropolitan University)

To prevent traffic accidents or mitigate their damage, it is necessary to investigate their causes. For this investigation, it is important to reconstruction the accident as reliably as possible. In this study, we studied a method for high-precision estimation of driving trajectories using drive recorder video and 3D-LiDAR data to enhance the reliability of accident reconstruction.

Study on Accident Reconstruction and Its Usefulness Using Combined Drive Recorder Video and EDR Data and with High-Precision Estimation Results of Vehicle Trajectory

Kotaro Ishizuka (Institute for Traffic Accident Research and Data Analysis)･Hideki Matsumura (NALTEC)･Kazuki Tainaka (Tokyo Metropolitan University)･Motoki Sugiyama (Institute for Traffic Accident Research and Data Analysis)･Naoyuki Kubota (Tokyo Metropolitan University)

Regarding accident reconstruction, which is important for investigating the causes of traffic accidents, the authors proposed an objective and quantitative method using EDR data and drive recorder video, as well as a high-precision driving trajectory estimation method using drive recorder video and 3D-LiDAR data. In this study, these methods were applied to reconstruct an actual accident and analyze it.