Performance-based evaluation method for detectability of alert signs on vehicle meters display

Tsukasa Kimura･Yurie Shin･Ryuraro Oe (Graduate School of Human Sciences,The University of Osaka)･Masanori Furuya･Tomomichi Uekuri (Nissan Motor)･Kazumitsu Shinohara (Graduate School of Human Sciences,The University of Osaka)

This study investigated vehicle meter display visibility and evaluation methods based on physical display characteristics and the drivers’ knowledge for meter display. The results indicated that visual complexity and location of information placement are critical factors affecting visibility. Furthermore, the results suggested that the proposed experimental method may be an effective method for evaluating the visibility of designed vehicle meters.

Influences of In-vehicle Information Presentation using Animation Display on Drivers’ Behaviors
-Discussion on Information Providing by V2X System on Intersections with Poor Visibility-

Akira Ohtani･Ryohei Homma (JARI)･Masaaki Abe (JAMA)

In this study, we investigated the influences on driving behaviors when animation information about the appearance of vulnerable drivers at intersections with poor visibility was presented to drivers. Based on the results of a driving simulator experiment, the influences of the animated display on driving behaviors and driver distraction were discussed.

Influence of humanoid robot interaction on pedestrian-aware driving behavior

Ayaka Togiya･Hirotaka Yamamoto (Kyoto Institute of Technology Graduate School)･Mariko Osaka (The University of Osaka)･Yukiko Nishizaki (Kyoto Institute of Technology)

With the aim of proposing a new way to use humanoid robots for safe driving support, we examined whether interaction with a robot before driving promotes drivers' empathetic driving behavior. Drivers play a game involving imitation of behavior with a humanoid robot before driving. We examined whether this would lead to differences in the degree to which drivers slow down to give priority to pedestrians at unsignalized crosswalks.

Basic Study on a Method for Estimating Arousal Level Focusing on Physiological Information of Drivers Conversing with Generative AI

Tatsuya Sato･Yuta Ogura･Shunta Takahashi･Komei Hayashi･Hidenobu Takao (Kanagawa Institute of Technology)･Ayumu Kawata･Yusuke Tanizawa･Hiroaki Hashimoto･Rusako Fujino･Nagata Hideki (Pioneer)

In the future, an interactive system that maintains arousal level by interacting with a generative AI is considered useful. To achieve this, it is necessary to estimate the driver's arousal level, but this method has not yet been generalized. In this study, we investigate a basic method for estimating the driver's arousal level based on physiological information using deep learning, especially in the low arousal level range.

Multi-Turn Dialogue with Large Language Model for In-Vehicle Spoken Dialogue Systems

Akinobu Lee (Nagoya Institute of Technology)･Koichiro Karasawa (Toyota Systems)･Atsunobu Kaminuma (International Professional University of Technology in Tokyo)

Dialogue systems based on large language models (LLMs) have been rapidly adopted for in-vehicle user interfaces. However, controlling the amount of information in a LLM-generated response through simple prompting is inherently difficult, thus potentially increases drivers' mental work load. This study examines incorporating several multi-turn, step-by-step strategies into LLM-based spoken dialogue system, aiming to keep user convenience.

Evaluation of a Human Machine Interface for Vehicle – infrastructure Cooperative Driver Assistance at Right Turns at Intersections

Kimihiko Nakano･Jun Sawada･Masaaki Onuki (The University of Tokyo)

This study investigates whether information provided from infrastructure to vehicles in cooperative Level 4 automated driving can also serve as effective driving support. Human-machine interfaces (HMIs) are proposed to convey the presence or absence of obstacles, or right-turn initiation decisions at intersections. Their effectiveness is evaluated through driving simulator experiments.

Evaluation of Driver Acceptance in in-Vehicle Driving Assistance Information Across Various Situations Using Event-Related Potentials

Jongseong Gwak (Takushoku University)･Hiroshi Yoshitake･Motoki Shino (Institute of Science Tokyo)

To evaluate driver acceptance of driving assistance information, we examined central nervous responses when suitable and unsuitable information was presented, focusing on differences across information types. Using a driving simulator and biosignal measurement system, we analyzed event-related potentials. Results showed that P300 peak amplitudes varied depending on the type of information under different suitability conditions.