OAK

Indoor Car VR Simulation Usability Evaluation: Automatic Generation of Car VR Contents with Visual Odometry

Metadata Downloads
Abstract
In line with the active research on autonomous driving, there is a significant amount of research being conducted on the use of indoor vehicle simulations utilizing motion simulators and virtual reality (VR) to enhance safety and efficiency. In in-car VR technology, where vehicles serve as a platform for VR content, synchronizing the motion of the vehicle with the virtual environment is done to reduce sickness and enhance immersion. Traditional sensor-based approaches for synchronizing vehicle motion with the virtual environment have disadvantages, leading to the exploration of vision-based methods. In this study, we present a system that enables car VR simulation in indoor motion simulators by implementing a visual odometry algorithm that estimates the motion of the vehicle using driving videos. The accuracy of the visual odometry was evaluated using the KITTI odometry driving dataset, and the system successfully reproduced car VR contents in indoor environments. User experiments were conducted for 16 participants to compare the user experience (UX) between simulations using ground truth and estimated data. The naturalness of motion in the estimated simulations was assessed through user perception tests. The impact of different driving conditions on UX when using estimated motion data was also investigated. The results indicate that vision-based methods offer a comparable level of UX, not falling short when compared to sensor-based approaches. Users perceived greater naturalness in straight driving conditions on smooth paths. This study contributes to understanding the potential of vision-based techniques in engaging car VR experiences.
Author(s)
Heesan Yang
Issued Date
2023
Type
Thesis
URI
https://scholar.gist.ac.kr/handle/local/19401
Alternative Author(s)
Heesan Yang
Department
대학원 융합기술학제학부(지능로봇프로그램)
Advisor
Kim, SeungJun
Degree
Master
Appears in Collections:
Department of AI Convergence > 3. Theses(Master)
공개 및 라이선스
  • 공개 구분공개
파일 목록
  • 관련 파일이 존재하지 않습니다.

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.