Preventing Imminent Collisions between Co-Located Users in HMD-Based VR in Non-Shared Scenarios

Authors: 
Allan Hanbury
Allan Hanbury
Type: 
Speech with proceedings
Proceedings: 
Proceedings of the 30 th International Conference on Computer Animation and Social Agents
Publisher: 
CASA 2017
Pages: 
37 - 46
Year: 
2017
ISBN: 
ISBN: 978-89-89453-82-6
Abstract: 
This paper presents two experiments set in a<br> multi-user HMD-based VR system where users<br> navigate by real walking in a large real and vir-<br> tual area. We investigate a case that could be<br> used in a multi-user VR game or a training ap-<br> plication: several users are walking in the same<br> physical space without seeing each other in the<br> virtual environment. Such a scenario involves<br> the risk of collisions between users. In the first<br> experiment, we investigate the strategy of stop-<br> ping a walking user in a dangerous situation. In<br> particular, we compare the effectiveness and the<br> perceived difficulty of two visual and two audi-<br> tory stopping signals. The results of this com-<br> parison show that the tested visual and auditory<br> signals are equally effective in stopping users.<br> With both visual and auditory signals, partici-<br> pants prefer the signal to contain a "stop" com-<br> mand. In the second experiment, avatars are<br> displayed at users´ positions if the distance be-<br> tween users is dangerously small. The method<br> is tested with four avatars of various degrees of<br> anthropomorphism and in two different appli-<br> cation scenarios. Our results suggest that the<br> type of scenario influences users´ preference of<br> a notification avatar. It is sufficient to display<br> an area occupied by other users in scenarios<br> with specific goals and interactive content. If<br> users are exploring a virtual world without hav-<br> ing any other goal, they prefer to see human-<br> like avatars as a possible collision notification.
TU Focus: 
Computational Science and Engineering
Reference: 

I. Podkosova, H. Kaufmann:
"Preventing Imminent Collisions between Co-Located Users in HMD-Based VR in Non-Shared Scenarios";
Vortrag: CASA 2017, Seoul, South Korea; 22.05.2017 - 24.05.2017; in: "Proceedings of the 30 th International Conference on Computer Animation and Social Agents", CASA 2017, (2017), ISBN: 978-89-89453-82-6; S. 37 - 46.

Zusätzliche Informationen

Last changed: 
21.02.2018 18:59:25
Accepted: 
Accepted
TU Id: 
265833
Invited: 
Department Focus: 
Media Informatics and Visual Computing
Author List: 
I. Podkosova, H. Kaufmann
Abstract German: 
This paper presents two experiments set in a<br> multi-user HMD-based VR system where users<br> navigate by real walking in a large real and vir-<br> tual area. We investigate a case that could be<br> used in a multi-user VR game or a training ap-<br> plication: several users are walking in the same<br> physical space without seeing each other in the<br> virtual environment. Such a scenario involves<br> the risk of collisions between users. In the first<br> experiment, we investigate the strategy of stop-<br> ping a walking user in a dangerous situation. In<br> particular, we compare the effectiveness and the<br> perceived difficulty of two visual and two audi-<br> tory stopping signals. The results of this com-<br> parison show that the tested visual and auditory<br> signals are equally effective in stopping users.<br> With both visual and auditory signals, partici-<br> pants prefer the signal to contain a "stop" com-<br> mand. In the second experiment, avatars are<br> displayed at users´ positions if the distance be-<br> tween users is dangerously small. The method<br> is tested with four avatars of various degrees of<br> anthropomorphism and in two different appli-<br> cation scenarios. Our results suggest that the<br> type of scenario influences users´ preference of<br> a notification avatar. It is sufficient to display<br> an area occupied by other users in scenarios<br> with specific goals and interactive content. If<br> users are exploring a virtual world without hav-<br> ing any other goal, they prefer to see human-<br> like avatars as a possible collision notification.