Feasibility Study of a Hazard Avoidance Brake Control System Using V2N Technology

Yoshiaki Irie (Toyota Motor)･Fumiaki Ise･Yukihiro Yamamoto (Technoco Corp.)･Keita Arita･Junki Ito (SoftBank Corp.)･Dror Elbaz･Tal Lavi (Eye-Net ltd.)

With the aim of enhancing ADAS capabilities by addressing hazard detection outside the driver’s field of view, this study explores the potential of V2N technology. By utilizing smartphones as external sensors, we examine the effectiveness of V2N in improving hazard detection, the positioning performance of smartphones, requirements for cloud systems, and the current challenges involved.

A Note on the Information Provision of Hazards in Urban Areas Caused by Roadside Snow Accumulation

Chinami Fukui (Graduate School of Engineering, Hokkaido University)･Sho Takahashi･Toru Hagiwara (Faculty of Engineering, Hokkaido University)･Hidekatsu Hamaoka (Graduate School of Engineering and Resource Science, Akita University)

In urban areas of snowy and cold regions during the winter, snow accumulation along the roadside not only reduces visibility but also diminishes the sensing performance of vehicles. This study examines the effectiveness of a new information provision system designed to alert drivers to hazardous locations. These hazardous locations are where snow piles resulting from snow removal may lead to an increased likelihood of overlooking other road users.

Initial study on position estimation of surrounding traffic participants observed by vehicle with position uncertainty

Kota Watanabe･Takuma Ito (The University of Tokyo)

For cooperative utilization of information about surrounding traffic participants observed by a vehicle, it is necessary to integrate two uncertainties: the uncertainty of the vehicle's position, and the uncertainty of the relative positions from the vehicle. In this study, we propose a method to integrate these two uncertainties and estimate absolute positions of surrounding traffic participants. In the proposed method, vehicle’s position and traffic participants' relative positions are estimated by Extended Kalman Filters, and then the two estimations are integrated with considering the uncertainties. Through simulation experiments, we validate the feasibility of the proposed method.

Initial study on method of extracting potentially dangerous spots to improve machine learning model for speed decision on community roads

Yuki Sadanaka･Keita Hori･Takuma Ito (The University of Tokyo)

Machine learning model for speed decision on community roads needs to adapt to the potential risk at blind spots. In this study, we aim at improving the machine learning model for speed decision on community roads so that the model can adapt to the potential risk. For this purpose, we develop a method of extracting potentially dangerous spots by using the machine learning model and pseudo careful driving data.

A proposal for indicators of proximity and safety margin for evaluating behavioral safety of automated vehicles

Yuichi Saito･Syohei Shimotori (University of Tsukuba)･Hideo Inoue (Kanagawa Institute of Technology)

A common understanding of what indicators and how to verify the safety of automated vehicles in response to possible hazardous events is lacking. In this paper, we propose an evaluation method for behavioral safety using a measure of proximity to other traffic participants and a measure of safety margin for accident avoidance in traffic conflict scenarios, and discuss its effectiveness.