Viewing Data Together — In VR!
Putting people in the same immersive space for data analysis.
In recent years, new technologies have enabled novel ways to support collaborative visual data analysis. However, most studies have focused on how groups utilize shared surface interfaces. In a study published by Benjamin Lee and others from Monash University at IEEE VIS 2020, the researchers sought to learn how people would tackle visualization tasks using a shared flexible virtual workspace called FIESTA: the Free-roaming Immersive Environment to Support Team-based Analysis. The study findings suggest that the choice to use 2D surfaces and 3D space depend on the type of visualisation used, and that participants tend to follow social protocols outside of tightly-coupled collaboration.
The work started with designing and developing FIESTA. The system offers a baseline level of data visualization functionality including 2D and 3D visualizations; grasping and ranged pointing interactions; an ability to create visualizations on a panel and in 3D space; and pointer, brushing, and annotation tools. FIESTA supports a free roaming shared environment using a tetherless VR setup to create a co-located immersive workspace. In this space, users are depicted as virtual avatars mapped to their real-world positions and labeled with unique name plates and colors. To ensure that users can utilize this space naturally, the researchers established a room metaphor with a four-wall room layout and a tabletop in the center. The wall allows visualizations to be “pinned” on their vertical surface, while the tabletop enables visualizations to “rest” on it.
The researchers conducted the study with 10 groups of 3 participants during 2 study parts with 5 groups each. In part A, participants were limited to only 2D surfaces, while those in part B got to experience 3D visualizations. The study used a housing auction result dataset from the Melbourne, Australia region, focusing on two types of data analysis tasks: directed and free exploration. The directed task includes specific questions related to the dataset, while the free exploration task asks participants to find interesting insights from the same dataset and present them. Both types of tasks required the groups to achieve their answers with consensus.
During each 90-minute sessions, each group was first trained on how to use FIESTA, then given the directed tasks, free exploration task, together with mid- and post-study questionnaires. A PC recorded the positions of the participants’ heads, hands, and other objects in the virtual room, as well as the interactions’ duration, and a top-down view of the virtual room.
The research team structured the findings using a representative group workflow, which include major stages in the groups’ workflows to address the visual analytics tasks. Each stage features a set of notable behaviors, with each group either following or deviating from the stage.
Using the representative group workflow approach, the researchers were able to draw crucial conclusions regarding how users tackle visual analytics problems in a shared virtual space. They found that collaborative visual analytics is positively encouraged by immersive environments. Users also willingly utilized 3D visualizations, but these call for intuitive, consistent UI designs, while the presence of 3D elements heavily influences how users organize them. Additionally, users only care about a virtual environment’s affordances if they offer a tangible benefit, as most groups chose to suspend the 3D visualizations freely in space instead of putting them on the virtual tabletop.
The findings also suggest that an equal distribution of interaction resources allows for both independent work and collaboration, with individual work spaces defined by each user’s movement patterns and placement of artifacts. Participants rarely interacted with objects created by others, but they could naturally transition to tightly-coupled collaboration either by observing each other or discussing, while forming a good balance between workspace awareness and privacy. However, ensuring proper perspective during collaboration with 3D visualizations is tricky due to different viewing angles.
Using FIESTA, researchers from Monash University made meaningful observations regarding behaviors that are unique to visual analytics in an immersive shared space. The study suggests that participants found the shared VR environment to be useful for both working independently and sharing findings with each other, that the panel metaphor of the virtual walls effectively supported 2D visualizations for presentation and analysis, and that 3D visualizations were often preferred to be freely suspended in convenient locations in space. The study also confirms results from previous studies on non-immersive environments regarding how groups would organically divide the shared environment into individual spaces, address tightly-coupled collaboration either by observing each others’ work or through discussion, and employ avatars and pointers to facilitate collaboration.
More Information
- B. Lee, X. Hu, M. Cordeil, A. Prouzeau, B. Jenny and T. Dwyer, “Shared Surfaces and Spaces: Collaborative Data Visualisation in a Co-located Immersive Environment.” IEEE Transactions on Visualization and Computer Graphics, vol. 27, no. 2, pp. 1171–1181, Feb. 2021, doi: 10.1109/TVCG.2020.3030450.
