Development of a Simulation to Estimate Emergency Transport Time and Its Simplified Method

Shinji Asakura･Heishiro Toyoda･Tomoyuki Miyoshi･Hiroto Kawano･Takashi Moriuchi･Shinji Yamagiwa･Maria Yasuda (Toyota Motor)

The emergency transport time has been increasing year by year, and there is a desire to reduce this transport time. One potential approach to study how we can shorten emergency transport times is to use traffic flow simulations. In this study, we developed a simulation model for emergency transport. Additionally, we developed a simplified estimation method to reduce the time needed to prepare the emergency driving model for future studies and to expand the options for routes being considered for emergency driving.

Analyzing the Discrepancy Between D-Call Net Based and Kinematics-Based Delta-V

Noboru Tanase･Shizue Katsumata･Takahiro Ando･Yasushi Nagaoka (Toyota Motor)･Mayu Ishii (Institute for Traffic Accident Research and Data Analysis)

D-Call Net constructs mortality risk curves based on pseudo ΔV calculated from accident data, while actual reports use the airbag ECU's ΔV to calculate mortality rates. Both ΔVs generally align in translational motion but may differ in rotational accidents. By matching accident data with D-Call reports, the differences in ΔV and their causes are analyzed.

Study on actual situations of D-Call Net by matching automatic notification data with ITARDA Macro data

Toru Kiuchi (ITARDA)･Nobuo Saito･Ichiro Ando (JAPAN MAYDAY SERVICE)･Mayu Ishii･Eiko Kagesawa (ITARDA)

In recent years, the number of automatic notifications has dramatically increased due to the spread of vehicles equipped with D-Call Net. The authors were able to obtain notification data from three new OEMs in addition to the existing one. Therefore, we conducted a new effect study by matching the most recent notification data from 2021 to 2023 with ITARDA macro data. Again, as in the previous study, the focus was on single-vehicle accidents.

Utilization of image recognition for pedestrian injury prediction considering vehicle collision areas.

Mie Tokuyama･Takahiro Andoh･Noboru Tanase･Shizue Katsumata (Toyota Motor)･Kohji Ichikawa (Toyota Technical Development)

To reduce traffic accident fatalities, early medical intervention is essential. This requires a higher accuracy in predicting human injuries. In this study, we will focus on pedestrians, who account for the highest number of traffic deaths. In addition to traditional collision speed, we will use the height of pedestrians and the vehicle collision areas measured by image recognition for the injury prediction algorithm.

Development of a Collision Detection Model Using Dashcam Audio Information (1st report)

Yuki Nomura･Shouhei Kunitomi･Yoshihiro Sukegawa (JARI)･Yasushi Nagaoka (JAMA)

This study developed a machine learning model for detecting collisions with passenger cars and vulnerable road users or motorcyclists using audio data recorded by dashcams. The model successfully detected 72 collisions out of 79 test data based on the audio data during the collisions. The analysis also discussed factors contributing to false detection and non-detection.