The Effects of Co-Present Embodiments on Awareness and Collaboration in Tabletop Groupware David Pinelle School of Computer Science University of Nevada, Las Vegas pinelle@cs.unlv.edu Miguel Nacenta, Carl Gutwin Department of Computer Science University of Saskatchewan {nacenta,gutwin}@cs.usask.ca Tadeusz Stach School of Computing Queen’s University tstach@cs.queensu.ca ABSTRACT Most current tabletop groupware systems use direct touch, where people manipulate objects by touching them with a pen or a fingertip. The use of people’s real arms and hands provides obvious awareness information, but workspace access is limited by the user's reach. Relative input techniques, where users manipulate a cursor rather than touching objects directly, allow users to reach all areas of the table. However, the only available awareness information comes from the virtual embodiment of the user (e.g., their cursor). This presents designers with a tradeoff: direct-touch techniques have advantages for group awareness; relative input techniques offer additional power but less awareness information. In this paper, we explore this tradeoff, and we explore the design space of virtual embodiments to determine whether factors such as size, realism, and visibility can improve awareness and coordination. We conducted a study in which seven groups carried out a picture-categorizing task using seven techniques: direct touch and relative input with six different virtual embodiments. Our results provide both valuable information to designers of tabletop groupware, and a number of new directions for future research. KEYWORDS: Tabletop groupware, embodiments, interaction techniques. INDEX TERMS: H.5.3 [Information Interfaces]: CSCW 1 INTRODUCTION Tabletop groupware systems allow co-present collaborators to work together over a shared horizontal display. Tables are a natural site for group work, both because of their ubiquity in the real world, and because their physical characteristics support coordination and communication, such as the face-to-face orientation of people around the table, the central location of work artifacts, and the use of direct touch to manipulate objects on the work surface. Direct touch – where people manipulate objects by touching them with a pen or a fingertip – provides a number of benefits for collaboration. In particular, the use of people’s real arms and hands provides awareness information about ‘who is working where’ on the table, and makes it easy to watch other people work [16]. However, direct touch also has disadvantages: it can be difficult for people to reach objects that are far away; arms and bodies can get in the way of each other, preventing people from working in the same space at the same time; and it can be awkward or uncomfortable to work close to another person. One way to deal with these problems is to use relative rather than direct input techniques – that is, techniques where each person manipulates a cursor rather than touching objects directly. Relative input techniques allow reaching to any part of the table and allow people to work in the same place, but since they do not use physical arms, this source of awareness information is lost. The only awareness information produced by a relative input technique comes from the virtual embodiment of the user (e.g., their cursor) on the table. This visible representation provides each user with feedback about their own actions, but as a side effect, also provides awareness to other members of the group. Although this is the same mechanism by which real arms convey awareness, virtual embodiments are much less obvious than physical arms. Tabletop designers are thus presented with a tradeoff: direct- touch techniques have advantages for group awareness but limitations for individual work; relative input techniques offer additional power but provide less awareness information. Most existing table systems have chosen direct touch input [1,4,14], but in fact little is known about how different techniques affect groupware usability. In this paper, we explore the tradeoff between individual power and group awareness, and investigate the design of virtual embodiments for tabletop groupware. We built several different tabletop embodiments, and studied their use in a realistic collaborative task. We were interested in three main questions: • What are the differences between physical and virtual embodiments in terms of collaboration factors such as awareness, distraction, territoriality, and movement patterns? • Does the appearance of a virtual embodiment (e.g., its size and shape) affect these factors, and can a virtual embodiment approach the awareness of a physical arm? • Overall, do users prefer direct touch and physical embodiment, or relative input and virtual embodiment? To investigate these questions, we carried out a study in which seven groups of three people carried out an open-ended picture- categorizing task on a digital table. We gathered movement data, tested people during the task on their awareness of other’s locations, and asked several survey questions about perceived awareness, distraction, social distance, and overall preference. The study provided some results that confirmed our expectations (e.g., people were significantly more accurate in guessing others’ locations with direct touch), but also many that were surprising. For example: • type of virtual embodiment had no effect on awareness accuracy – even though some of the embodiments were much larger and more obvious than others; • virtual embodiments did not appear to carry the same social constraints as did direct touch: people were more likely to cross embodiments, and they were much more likely to move into another user’s personal territory; • although there were few differences between the virtual embodiments in objective measures, there were strong subjective differences; in general, people preferred larger and more realistic embodiments, and felt that these provided much better awareness;