CHI
Doctorial Consortium
changing the world, changing ourselves
Multiple Perspectives for Collaborative Navigation in CVE Huahai Yang School of Information, University of Michigan Ann Arbor, MI 48109, USA
[email protected] Abstract Drawn from empirical studies on spatial cognition, this work explores ways of dynamically integrating others’ perspectives and incorporating different views into a single interface for a 3D CVE user. It also designs an empirical study to test the effectiveness of different perspective displays on collaborative navigation performance.
throughout the human development process [2]. Although it is difficult, some findings suggest that certain prior movement activities in the environment could improve the understanding of other perspectives than one’s own [9]. This points us to a direction of improving spatial understanding by dynamically presenting the transition of viewpoints to users.
Keywords CVE, multiple-perspective, awareness, navigation
Effect of Different Perspectives on Navigational Performance Performance of navigational tasks often depends upon comparisons of information expressed in different spatial perspectives. For example, the navigator’s forward field of views is generally expressed in a highly egocentric “immersed” perspective, while maps and goals are often expressed in a more exocentric world perspective [10]. It appears that in a real world navigational task, which involves both goal finding and locomotion sub-tasks, different perspectives of varied egocentricity are required in the different stages of the process. It would be well worth trying to combine these different perspectives in a meaningful way so that the benefit of each perspective can be optimized while cost minimized. Several attempts have been made to combine multipleperspectives to facilitate navigation. Tlauka et al. show an improved navigational performance using an egocentric and exocentric dual perspectives display [7]. Ramloll & Mowat use clickable snapshots of a location to construct a physical spatial cognitive map representation, which proved to decrease user’s workload. [6] These promising results strengthen our belief that smooth transitions among different perspectives enhance navigational performance.
INTRODUCTION A spatial 3D collaborative virtual environment (CVE) system allows multiple users who are distributed across a computer network to enter a shared virtual environment constructed from 3D computer graphics. Each participant has a visual embodiment called an avatar. Users can manipulate objects in the virtual environment and interact with others as well [1]. Many tasks in these systems have the element of collaborative navigation, in which several users explore the environment independently, then try to navigate to a common location and work together on some objects in the environment. In these tasks, one of the prominent problems is the difficulty of establishing mutual awareness among CVE participants [4]. It is found that the reference to an object and its location became very difficult because of “tunnel vision” and misunderstanding of others’ perspectives. Participants have to compensate this loss of referent with spoken accounts of their actions, which greatly increased the communication overhead. In order to facilitate this type of collaborative work, we need to study people’s understanding of multiple spatial perspectives in collaboration. It is also important to devise tools that could enhance users’ awareness of others’ perspective and improve their overall spatial awareness. RELATED WORK IN THE EMPIRICAL STUDIES A number of empirical studies in the areas of spatial cognition could shed light on our understanding of the processes involved in a multiple perspective navigational task in CVE, and could inspire us to devise new interaction techniques to facilitate this type of task. Spatial Perspective Taking It is well established that the ability to imagine another’s perspective is a rather difficult and emerging cognitive skill Copyright is held by the author/owner(s). CHI 2002, April 20-25, 2002, Minneapolis, Minnesota, USA. ACM 1-58113-454-1/02/0004.
560
Language Communication of Spatial Perspectives When participants carry on a spatial navigational task in CVE, they are very likely to engage in some verbal communication tasks in the mean time. There has been evidence showing that the availability of verbal channel affects performance on spatial cognition task [3]. When verbal communication is carried out, spatial task performance could be decreased. It is both theoretically and practically useful to see how this effect plays out in a CVE collaborative navigational task. Besides the possible effect on spatial performance and communication overhead, another possible effect is on verbal perspective taking of the speakers. It is reported that in a conversation, a speaker takes the information-seeker’s spatial perspective whenever possible. When the mental workload is high however, they will take a neutral or their own perspective.
minneapolis, minnesota, usa • 20-25 april 2002
[8] These effects could be used to measure the degree of success of a navigational tool in term of facilitating collaborative navigational task in CVEs. PROPOSED WORK A scenario is designed to empirically test the effect of different multiple perspective viewing displays on a collaborative navigation task. In this scenario, a pair of participants will take part in a collaborative version of the visualization task presented in McCormick et al. [5] One participant (navigator) will be responsible for identifying and communicating the location of a target to the other (driver), while the driver is traveling to the target without prior knowing where target is located. The driver will always use an immersed first-person (egocentric) perspective, while we manipulate the type of the navigator’s perspective displays. The effects of different perspective displays on navigator’s direction-giving ability will be our performance measurement. The navigator’s perspective displays are updated as the driver moves. For example, one type of display is first-person view, where the navigator sees exactly what the driver sees; another type is third-person view, where the navigator sees the driver as an avatar moving in the space. Prototype System Design: M-SCOPE Based on the findings from the literature, it is apparent that different perspective displays are suitable for different phase of the navigational process as a whole; hence a flexible perspective control tool is needed. Also, it appears to be desirable to see other people’s perspective in order to better support collaborative navigation. The proposed work will investigate ways to make these desirable features possible in a prototype CVE system. By ways of meaningful (for the users) camera control, we will achieve seamless integration of maps and scenes, and smooth transition between users’ perspectives. A multiple-perspective small-group collaborative operating environment (M-SCOPE) is in development to implement the abovementioned features. Working with Java3D and JSDT, M-SCOPE library serves as a middle layer between 3D renderer (Java3D) and network middleware (JSDT). In M-SCOPE, the users’ viewpoint information (size, locations and orientations) is synchronously disseminated to all participants. This viewpoint information will be used to construct flexible perspective displays that enhance users’ awareness of others’ perspectives and improve their overall spatial awareness. Empirical Study In the field of aviation, where most available empirical data concerning the effect of multi-perspective on navigation come from, the problem of multi-perspective often refers to as the problem of Frame of References (FOR) [10]. The degree of egocentrism in reference is a continuous variable that goes from the lowest as in a fixed orientation ‘bird’s eye’ perspective to the highest as in an ‘immersive first
Doctorial Consortium
person’ perspective. The main focus of the study is to test how different FORs affect navigator’s ability to guide the driver. The hypotheses: the more egocentric FORs will allow navigator to give better instruction to driver in term of travel, but will be slow in giving such instruction since it takes more time to find the target with egocentric FOR; on the contrary, the more exocentric FORs will allow faster instruction onset, but the communication overhead will be higher since it is difficult to communicate the location of target when navigator and driver have different perspectives; our M-SCOPE integrated multiple perspective approach will perform better than any of the other single or combined perspective displays in term of time spent accomplishing the task and communication overhead. ACKNOWLEDGMENTS I thank Gary M. Olson for excellent guidance of this thesis work. REFERENCES 1. Churchill, E. F., Snowdon, D. N., & Munro, A. J. (2001). Collaborative virtual environments: digital places and spaces for interaction. London: Springer. 2. Cox, M. V. (1978). Order of the acquisition of perspective-taking skills. Developmental Psychology, 14(4), 421-422. 3. Hermer-Vazquez, L., Spelke, E. S., & Katsnelson, A. S. (1999). Sources of flexibility in human cognition: Dualtask studies of space and language. Cognitive Psychology, 39(1), 3-36. 4. Hindmarsh, J., Fraser, M., Heath, C., Benford, S., & Greenhalgh, C. (1998). Fragmented Interaction: Establishing Mutual Orientation in Virtual Environments, in Proceedings of ACM CSCW’98 (Seattle WA, November 1998), ACM Press, 217-226. 5. McCormick, E. P., Wickens, C. D., Banks, R., & Yeh, M. (1998). Frame of reference effects on scientific visualization subtasks. Human Factors, 40(3), 443-451 6. Ramloll, R. & Mowat, D. (2001). Wayfinding in Virtual Environments using an Interactive Spatial Cognitive Map, in IV2001 Proceedings (London), IEEE Press, 574-583. 7. Tlauka, M., Stanton, D., & McKenna, F. P. (2000). Dual displays. Ergonomics, 43(6), 764-770. 8. Schober, M. F. (1993). Spatial perspective-taking in conversation. Cognition, 47(1), 1-24. 9. Weatherford, D. L., & Cohen, R. (1980). Influence of prior activity on perspective taking. Developmental Psychology, 16(3), 239-240. 10. Wickens, C. D. (1999). Frames of reference for navigation. In D. Gopher & A. Koriat (Eds.), Attention and Performance XVII (Vol. 17). Cambridge, MA: MIT Press, 113-144.
561