In Virtual Reality, having a virtual body opens a wide range of possibilities as the participant's avatar can appear to be quite different from oneself for the sake of the targeted application (e.g., for perspective-taking). In addition, the system can partially manipulate the displayed avatar movement through some distortion to make the overall experience more enjoyable and effective (e.g., training, exercising, rehabilitation). Despite its potential, an excessive distortion may become noticeable and break the feeling of being embodied into the avatar. Past researches have shown that individuals have a relatively high tolerance to movement distortions and a great variability of individual sensitivities to distortions. In this article, we propose a method taking advantage of Reinforcement Learning (RL) to efficiently identify the magnitude of the maximum distortion that does not get noticed by an individual (further noted the detection threshold). We show through a controlled experiment with subjects that the RL method finds a more robust detection threshold compared to the adaptive staircase method, i.e., it is more able to prevent subjects from detecting the distortion when its amplitude is equal or below the threshold. Finally, the associated majority voting system makes the RL method able to handle more noise within the forced choices input than adaptive staircase. This last feature is essential for future use with physiological signals as these latter are even more susceptible to noise. It would then allow to calibrate embodiment individually to increase the effectiveness of the proposed interactions.

When placed inside an immersive virtual simulation, subjects will tend to experience the feeling of being ’really there’ and to respond realistically to their environment, forgetting that it is not real. This behaviour is observed when subjects experience a high sense of presence, the sensation of being in a real place and that the scenario being depicted to them is real. Here we present an experiment designed to evaluate the impact of different levels of immersion, and of different blending of virtual and real objects and body representations, on participant’s subjective experience. Presence is evaluated with an innovative method combining the random introduction of breaks-in-presence (BiP) with a rapid decision-making test. Results show that the level of immersion impacts both the Sense of Presence (SoP) and the Sense of Embodiment (SoE), that the BiP has a limited impact on the SoE without breaking it, and that the level of confidence in the decision test correlates with both the SoP and the SoE.

Performing motor tasks in virtual environments is best achieved with motion capture and animation of a 3D character that participants control in real time and perceive as being their avatar in the virtual environment. A strong Sense of Embodiment (SoE) for the virtual body not only relies on the feeling that the virtual body is their own (body ownership), but also that the virtual body moves in the world according to their will and replicates precisely their body movement (sense of agency). Within that frame of mind our specific aim is to demonstrate that the avatar can even be programmed to be better at executing a given task or to perform a movement that is normally difficult or impossible to execute precisely by the user. More specifically, our experimental task consists in asking subjects to follow with the hand a target that is animated using non-biological motion; the unnatural nature of the movement leads to systematic errors by the subjects. The challenge here is to introduce a subtle distortion between the position of the real hand and the position of the virtual hand, so that the virtual hand succeeds in performing the task while still letting subjects believe they are fully in control. Results of two experiments (N=16) show that our implementation of a distortion function, that we name the attraction well, successfully led participants to report being in control of the movement (agency) and being embodied in the avatar (body ownership) even when the distortion was above a threshold that they can detect. Furthermore, a progressive introduction of the distortion (starting without help, and introducing distortion on the go) could even further increase its acceptance.

Hands deserve particular attention in virtual reality (VR) applications because they represent our primary means for interacting with the environment. Although marker-based motion capture with inverse kinematics works adequately for full body tracking, it is less reliable for small body parts such as hands and fingers which are often occluded when captured optically, thus leading VR professionals to rely on additional systems (e.g. inertial trackers). We present a machine learning pipeline to track hands and fingers using solely a motion capture system based on cameras and active markers. Our finger animation is performed by a predictive model based on neural networks trained on a movements dataset acquired from several subjects with a complementary capture system. We employ a two-stage pipeline, which first resolves occlusions, and then recovers all joint transformations. We show that our method compares favorably to inverse kinematics by inferring automatically the constraints from the data, provides a natural reconstruction of postures, and handles occlusions better than three proposed baselines.

Hands in virtual reality applications represent our primary means for interacting with the environment. Although marker-based motion capture with inverse kinematics (IK) works for body tracking, it is less reliable for fingers often occluded when captured with cameras. Many computer vision and virtual reality applications circumvent the problem by using an additional system (e.g. inertial trackers). We explore an alternative solution that tracks hands and fingers using solely a motion capture system based on cameras and active markers with machine learning techniques. Our animation of fingers is performed by a predictive model based on neural networks, which is trained on a movements dataset acquired from several subjects with a complementary capture system (inertial). The system is as efficient as a traditional IK algorithm, provides a natural reconstruction of postures, and handles occlusions.

We present a first study where we combine two asymetric virtual reality systems for telecollaboration purposes: a CAVE system and a head-mounted display (HMD), using a server-client type architecture. Experiments on a puzzle game in limited time, alone and in collaboration, show that combining asymetric systems reduces cognitive load. Moreover, the participants reported preferring working in collaboration and showed to be more efficient in collaboration. These results provide insights in combining several low cost HMDs with a unique expensive CAVE.

This project investigates the impact of a virtual zero gravity experience on the human gravity model. In the planned experiment, subjects are immersed with HMD and full body motion capture in a virtual world exhibiting either normal gravity or the apparent absence of gravity (i.e. body and objects floating in space). The study evaluates changes in the subjects' gravity model by observing changes on motor planning of actions dependent on gravity. Our goal is to demonstrate that a virtual reality exposure can induce some modifications to the humans internal gravity model, analogous to those resulting from real exposure (e.g. parabolic flights), even if users remain under normal gravity condition in reality.