Perceived physical presence in Mixed reality embodiment vs Augmented reality robot interaction

Abstract: This thesis presents a novel interaction model using mixed reality simulating a robot human interaction; a clay embodiment is overlaid with animated facial features using mobile augmented reality. One of the challenges when building a social agent, whether it is for education or solely social interaction, is to achieve social presence. One way to increase the feeling of presence is to have the agent physically embodied by using a robot. Earlier research has found that users listen more to robots that are present, than robots that are presented through a screen. But there are problems that come with robots that are not yet solved. Robot are expensive, they break, they are hard to update and they are very limited to the realm and problems they are built for: even standing up can be a challenge. This thesis tests if the theory of embodiment as a tool to heighten presence can be achieved, even if the robot and the interaction is only present in a screen. The clay embodiment is built by hand and later 3D scanned. The clay embodiment is tracked using Vuforia’s object recognition of the scan and is given an animatable face in a mixed reality setting through unity. The interaction of comparison and the basis of evaluation consist of a fully virtual robot head placed in 3D space using ground plane tracking. These interactions are compared separately and test subjects are only exposed to one type of interaction. Through the study the participants interacting with the clay embodiment rated the exeprience higher in respect to physical presences and scored better ability to recall details than the one with the fully augmented robot human interaction. The results were significant and indicate, with the reservation of false positives given the small participation sample, that mobile augmented reality agent interactions are improved, in respect to attention allocation and physical presence, by the use of mixed reality embodiments. Overall the interaction was very well perceived. Both conditions were highly enjoyed and critique mostly focused on the lack of complexity in the dialogue - the participants wanted more. Initial positive feedback states that this can and should be tested further.