Conclusion and Future Research

Demonstrating closely-coupled collaboration, as done in this thesis, is an important step towards an application allowing for many human interactions. Chapter 1 defined closely-coupled collaboration as a close coupling between object manipulation and human interaction, whereas the action of collaborating people is directly depending on each other. Margery [1999] categorised it as level 1 - co-existence and shared-perception; level 2 - individual modification of the scene; and level 3 - simultaneous interactions with an object. This work extended and clarified level 3 by highlighting the distinction between sequential and concurrent sharing of the same and different object attributes. Before natural and collaborative object manipulation can be applied in practical applications, developers and system designers need to understand how to support effective closely-coupled collaboration. The answer is not trivial and depends on a multitude of factors, a number of which were examined in this thesis.

While people cooperate with other people through an object, they use a variety of communicational resources to demonstrate their opinions, intention and needs to others. Verbal and non-verbal communication is often mutually supportive and the meaning of one can be changed or even lost without the other. This may be manifested, for example, as gesturing and posture to reinforce the emotion of the spoken word, or by talk and gesture guiding collaboration during shared object manipulation. When interacting remotely, supporting these forms of social human communication (SHC) can improve the effectiveness of collaboration. Chapter 2 introduced and discussed primary elements of SHC as well as measurement and need for a feeling of presence and co-presence to support distributed collaboration. It presented that the naturalness of collaboration depends on how well the object manipulation and forms of SHC are supported and mediated through tele-collaboration technology. However, as a survey in chapter 3 discussed, most technologies have difficulties in supporting natural object interaction with distributed groups while immersive systems seam currently to be best suited for such tasks. The survey suggests that, as of today, a system that allows users to share a common virtual space and to “step-into each others world” (Chapter 3, Figure 3-12c), such as an immersive CVE, provides the closest resemblance of co-location from distributed sites. In a CVE, remote people and shared objects can be situated in a shared synthetic environment, in which one can navigate around and interact with a computer-generated representation of objects and other participants. Thus, whereas video-conferencing systems allow people to look into each other’s space, CVEs allow people and data to be situated in a shared spatial and social context.

To demonstrate and evaluate distributed collaboration the Virtual Gazebo benchmark application was created. This contains a structured task that requires at least two users to collaborate closely with each other. Communication in this virtual environment is just as important for success as technological support for human interaction and collaboration. The benchmark’s successes and failures during development were documented in chapter 4 and the following chapters 5 to 7 examined the benchmark application in various user trials. This work demonstrated, within the confines of the application, that:

• distributed closely-coupled collaboration (including object focussed non-verbal communication and concurrent shared manipulation of objects)  is possible with today’s CVE technology  (Chapter 5&6)
• a difference in level of immersion between users of linked displays leads to the more immersed user taking a dominant role, as found by  Steed et al. [1999], held true for closely coupled collaboration (Chapter 5)
• however, the scale of impact depended on the closeness of collaboration, for example the method of shared object manipulation (Chapter 5)
• CVEs connected with immersive CAVE-like displays are well suited to support closely-coupled tasks (Chapter 6)
• using immersive displays supports a greater level of fluency in workflow (Chapter 6)
• People believe they are performing better in a “step-into” CAVE-like display than on a “look-into” desktop or “reach-into” displays. However, objective measurements of task performance indicate such improvements in object focussed collaboration but not for single user object interaction (Chapter 7)
• Although immersion seems to improve the perception and performance of collaboration, there are a number of factors that distract from the experience and still create a significant gap between the co-located and remote meeting. (Chapter 7&8), these and other supportive factors were summarised in a framework in Chapter 8 (Figure 8-1):
  • Some of these factors are in common with single user object interaction, for example immersion to create a feeling of presence or user interface for natural object manipulation
  • Others are more related to collaboration, like the requirements that SHC place on intuitiveness of medium and interface (may in light of limitations in both), placing a need on the simulation to compensate. For example, adding a visual signal to substitute for the feeling of touch when an object is selected (see Virtual Gazebo in Chapter 5&6) or bending simulation time through space to hide the network delay [Sharkey et al., 1998].
• A further outcome of this study is that many issues should be addressed to improve performance, handling and workflow of distributed collaboration. Therefore, the following section will discuss some future research directions that research community could take, to reach the goal of using collaborative VR in a wider range of activities.

Directions for Future Research

The research of this thesis focused mainly on the support for distributed closely-coupled collaboration using immersive display technology. While it showed that IPTs are useful for such collaboration, the research also raised some questions which could be answered in future research. For example, a major issue was the intuitiveness of the interface which resulted in issues while grasping objects and moving about within the immersive display. An alternative wireless interface with more precise tracking and a support for automatic constraint recognition could improve the ease of natural object interaction and lower the awareness of the technology. Some research in this area has been done and showed promising results [Marcelino et al., 2003; Osawa, 2006].

A further research direction might improve support for non-verbal communication through better gesture, posture and gaze support. The typical tracking interface allows only two to three sensors, to track for head and hand(s), which in combination with basic inverse kinematics give basic gesture support. Observations showed that participants made active use of this. However, more body sensors combined with enhanced avatar animation and support for facial expression could improve interactive communication and awareness of interactions from collaborators. Alternatively, a solution using video data of the actual user mapped onto the avatar could greatly enhance faithfulness and gain of trust.

Another problem of object interaction in immersive VR is adding the sense of touch through haptics interfaces. Although the research community for such devices is large, there are currently few haptic devices that would fit into a CAVE-like display. Furthermore, they can easily be seen which could detract from the illusion of presence.

A closer look into cultural differences between distributed groups, would further add to the understanding of how interfaces and application design have to be improved for global multicultural collaboration. In addition, most research is usually short term and a long term study would be beneficial to understand side effects of long term VR exposure as well as to identifying methods of adaptation (to technology, application and communication) used by participants.

Immersive CVEs appear promising for the support of distributed group collaboration and seem to be a good choice for studying people’s behaviour during such collaboration, something demonstrated by this work.  Applying the knowledge of the Virtual Gazebo research and related studies to create an application which could be used outside the academic research is desirable.

In final Summary

In an increasingly global economy there is increasing pressure to expand collaboration from co-located to geographically distributed groups, and this work has contributed to the understanding and further development of distributed closely-coupled collaboration, through immersive collaborative environments. This was approached primarily by measuring the impact of immersion and by isolating some key factors within this that have an impact on perceived and actual task performance. This work is contributing into the future development of distributed of collaboration by examining the strength and weaknesses of the technology supporting natural object focussed collaboration in distributed virtual environments and it is hoped that this work will generate a continued interest in many aspects of its content.