ANALYSIS AND COMPARISON OF SPEECH-BASED TELEPORTATION TECHNIQUES FOR IMMERSIVE VIRTUAL REALITY

Virtual Reality (VR) has significantly emerged in recent years in a large variety of fields, from the purpose of developing videogames to that of training, in which nowadays many companies are investing a huge amount of money. With the technology progress, Head-Mounted Displays (HMDs) have been created in order to cover the whole user’s Field of View (FOV) and give him/her a real sense of immersivity in the Virtual Environment (VE). Moreover, several hardware devices have been introduced to allow the user to perceive an increasing degree of immersivity while he/she is operating in VR, like haptic gloves for tactile sensations and force feedback returned by the virtual objects present in the VE. However, many limitations are intrinsic to VR itself. For example, while for small areas, it is possible, thanks to Room Scale VR, to detect and then reproduce the natural movements of the user as far as his/her translations and rotations are concerned, in a very large VE, like a hangar or a football field, it is not possible to provide him/her with a completely realistic interaction. In fact, since the user’s operating room is not so large, he/she necessarily has to find another way to navigate the virtual space in order to reach the desired places. When considering locomotion in VR, several issues must be considered in order to give the user a comfortable experience. About that, it is important to reduce as much as possible the phenomenon of motion-sickness; for example, the classical technique of using the controller’s analog stick in order to move toward a specific direction in space creates a discrepancy between the visual feedback and the feedback that is given by motion, producing nausea. For this reason, many alternatives have been studied during the years in order to avoid this problem. The one that is mostly used is teleportation, which allows to instantly change the Point of View (POV) of the user without emphasizing the movement. However, this kind of operation requires a way to indicate the Point of Interest (POI). In parallel with this, a way for the user to issue his/her commands through speech has been introduced in many VR applications for general interaction with the surrounding environment, for example as alternatives to menus and to manipulate objects. In this study, three teleportation techniques derived from literature have been implemented, in order to understand which one is preferable according to several metrics. In particular, the research focused on techniques based on speech. The first technique exploits the use of speech alone, requesting the user to utter phrases like “Take me to the table” in order to pass the system the intent and a semantic description of the POI. The second technique uses speech and head gaze together, by letting the user give a spatial information with the latter and triggering the action by using voice; the user can only say, for example, “Take me there”. Finally, the last technique combines the two approaches presented above in order to give the system both spatial and semantic information and allow the user to take advantage from both. Several tests with end-users were organized by using Unity as game engine and Microsoft Speech to recognize the voice commands. An ad-hoc environment was built in order to give the users the possibility to experiment the possible limitations of the three techniques, as well as their advantages.