Interaction technique’s and task’s degrees of freedom manipulation during orientation task in VR environments Manuel Veit * LSIIT-ULP (CNRS UMR 7005) Antonio Capobianco † LSIIT-ULP (CNRS UMR 7005) Dominique Bechmann ‡ LSIIT-ULP (CNRS UMR 7005) ABSTRACT In the present paper, we present our study about user’s behaviour during orientation tasks. More specifically, in this study we were interested in the influence of interaction modality and task’s com- plexity on user’s strategy to achieve orientation tasks. Because common Human - Computer Interaction (HCI) measure aren’t suf- ficient to understand user’s behaviour during the task, we introduce a new measure called IMDOFM, Independency’s Measure of De- grees Of Freedom Manipulation. This measure gives quantitative informations on how the user is orientating an object. Our results suggest that interaction techniques have a strong influence on the user’s strategy to rotate an object. It seems that even the interaction technique allows the users to integrate the manipulation of several task’s DOF in a single gesture, he tries to decompose the DOF ma- nipulation, thus dividing 3 DOF orientation tasks into successive 1 DOF tasks. On the contrary of what is usually observed, our re- sults show that this decomposition does not introduce any loss in precision or achievement time. Index Terms: H.5.1 [INFORMATION INTERFACES AND PRESENTATION]: Artificial, augmented, and virtual realities— [H.5.2]: INFORMATION INTERFACES AND PRESENTATION—User Interfaces 1 I NTRODUCTION Virtual Reality (VR) technology and its applications have been con- sidered as a major step towards a more direct and intuitive way of interacting with 3D scenes. By copying real world interactions into virtual applications, researchers were expecting to combine the best of them. However, it seems that interacting with a virtual envi- ronment needs to reconsider the way of designing interaction tech- niques even for the most simple actions like designating or rotating objects. Indeed, the interaction context of classical desktop stations is very different from what users may experience in VR environments. Such environments provide users with 3D stereoscopic displays im- mersing them in interactive virtual scenes (see Figure 1). Augment- ing interaction dimensionality by the addition of this third dimen- sion was considered as an interesting way of introducing more nat- ural interaction. As a matter of fact in VR environments users can interact with its surrounding scene using everyday commands like grabbing, turning, etc. [2] However while interaction seems more intuitive, VR setup are mainly used for visualisation applications. It seems that providing users with more natural commands is not sufficient to bring them with precise interaction techniques. Since the emergence of desktop environment in the late 70s, many efforts was made by HCI community to understand user’s behaviour when performing critical tasks like pointing or dragging. * e-mail:veit@lsiit.u-strasbg.fr † e-mail:capobianco@lsiit.u-strasbg.fr ‡ e-mail:bechmann@lsiit.u-strasbg.fr Figure 1: User is grabbing the object with one hand while performing a deformation with the other hand. Fundamental works like the definition of Fitt’s Law [4] helps de- signers to understand the user’s behaviour when pointing objects in desktop environments. Understanding user’s behaviour is critical to help designers developing efficient interaction techniques. Be- cause of fundamental differences in interaction context, theoretical framework already existing for desktop environments can not be di- rectly applied to 3D VR environments. The difficulties researchers encounters designing a generalisation of Fitt’s Law suitable for VR environments [6], illustrate difficulties that arise with such environ- ments. In this study, we are interested in understanding users’ behaviour during orientation task in semi-immersive virtual environments (SIVE). We studied users’ behaviour during orientation tasks using two different techniques, one integrating and the other facilitating the manipulation of task’s DOF. Even if each technique allow users to rotate virtual objects, their design is fundamentally different. By trying to understand the user’s behaviour during orientation tasks, we hope that it will help us design more efficient interaction tech- niques to rotate objects in SIVE. 2 PREVIOUS WORK The mental representation and the achievement of a rotation task is very complex [13]. Some studies have tried to investigate which factors, in addition to task’s complexity and interaction technique, influences task’s difficulty [11, 16]. LaViola et al. highlighted the fact that the presence of stereoscopy or head-tracking has a strong influence on performances [11]. Another study conducted by Prab- hat et al. [16] showed that the environment has a significant in- fluence on user’s performance during orientation tasks. It seems that performing the task in a fully immersive environment, like the CAVE, leads to longer achievement times and more important error rates than in a desktop environment. Even before the emergence of VR, rotation tasks in desktop en-