Sonification-based Steering Assistance for Curve Driving: Evaluations in Virtual and Real-World Environments (ICDA2025)

273 Views

July 03, 25

スライド概要

Sonification-based Steering Assistance for Curve Driving: Evaluations in Virtual and Real-World Environments
ICAD2025

profile-image

明治大学 総合数理学部 先端メディアサイエンス学科 中村聡史研究室

シェア

またはPlayer版

埋め込む »CMSなどでJSが使えない場合

ダウンロード

関連スライド

各ページのテキスト
1.

Sonification-based Steering Assistance for Curve Driving: Evaluations in Virtual and Real-World Environments Sayuri Matsuda, Satoshi Nakamura, Kento Watanabe (Meiji Univ.) Takeshi Torii, Hideyuki Takao, Yuki Mizuhara, Saeri Shimizu (SUBARU)

2.

Background • Novice drivers often struggle to drive through curves • Objective – To realize a practice method that helps novice drivers become better

3.

Measuring Good Driving Steering Corrections n 1 t Steering Corrections = 1 Good Fewer steering corrections Steering angle Steering angle A steering correction is defined as a change in the direction of the steering wheel's rotation speed. 3 5 1 7 9 2 4 6 8 Steering Corrections = 9 Bad More frequent steering corrections t

4.

Our Sonification Method • Our method generates a musical scale in real time, corresponding to the steering angle. [Matsuda 2022] – This method allows drivers to perceive the steering angle through sonification.

5.

Good vs. Bad Driving Example Good Fewer steering corrections Bad More frequent steering corrections

6.

Two types of Prototype System • Virtual Environment – Assetto Corsa • Real-World Environment – Actual vehicle

7.

Prototype for Virtual Env.

8.

Prototype for Real World Env. First Author (Sayuri Matsuda)

9.

Course (Virtual and Real) • In each environment, we used Automotive Business SUBARU R&E Center (SKC) – Length: 1800m (2-3 minutes per lap) – Virtual one was generated from the real.

10.

Experiment 1: Virtual Env. • N=26 (between subjects) • Comparative Methods – Sonification (Our Method) – No feedback (Baseline Method) • Procedure • Main Measure – Reduction in the number of steering corrections from pre- to final-exam

11.

Experiment 2: Real-World Env. • N=10 (between subjects) • Comparative Methods – Sonification (Our Method) – Instructor (An instructor sits next to the driver and provides guidance) • Procedure • Main Measure – Reduction in the number of steering corrections from pre- to final-exam

12.

Results: Experiment 1 (Virtual) • Participants paid more attention to the course with the baseline method than with ours. • Reduction ratio of steering corrections Our method > Baseline (No feedback)

13.

Results: Experiment 2 (Real) • Reduction ratio of steering corrections – No difference • Our system was effective for low-skill drivers, but not for high-skill drivers. • Some felt pressure from the instructor. Proposed: Effective Proposed: Ineffective Instruction Pressure?

14.

Discussion • Positive – Our method was effective to reduce steering corrections in virtual environment. – Our method performed as well as human instructors in real-world environment. • Some novice drivers get frustrated when being instructed, especially by their fathers. ➔ This method works better for them. • Negative – An experienced driver said sonification conflicted with engine sounds used to manage speed.

15.

Conclusion • Steering-angle sonification may help novice drivers improve without an instructor. • Future work – Apply our method to car parking situation – Present guide sound before the next curve