Adapting and tiling the streaming of virtual reality (VR) video content has the potential to reduce the ultra-high bandwidth requirements of this type of multimedia services. Towards that goal, the optimization of a number of aspects is currently actively being researched. Novel rate adaptation heuristics, sophisticated viewport prediction algorithms and streaming protocol optimizations have proven their value to improve certain aspect of the VR streaming chain. However, the interplay between all these different optimizations as well as their tradeoff has not yet been explored in an experimental playground. The purpose of this demonstrator is to provide a full end-to-end adaptive tile-based VR video streaming system where each of the optimization aspects can be tuned with and their effect illustrated on-site.

Virtual Reality (VR) has the potential to change not only to the way we consume and perceive entertainment but also to improve other important areas of society. One sector that is starting to benefit from the advantages of VR is the treatment of stress related mental illnesses. VR is able to bring relaxation therapy to the next level in which solutions can be scalable (without the need for real-time dedicated professionals) and personalized. This paper presents VRelax, a personalized VR relaxation therapy approach. By means of semantic methodologies and online learning techniques, VRelax provides a personalized, relaxing virtual environment to the user.

Online learning is the preferred option for professional training, e.g. Industry 4.0 or e-health, because it is more cost efficient than on-site organisation of realistic training sessions. However, current online learning technologies are limited in terms of personalisation, interactivity and immersiveness that are required by applications such as surgery and pilot training. Virtual Reality (VR) technologies have the potential to overcome these limitations. However, due to its early stage of research, VR requires significant improvements to fully unlock its potential. The focus of this PhD is to tackle research challenges to enable VR for online training in three dimensions: (1) dynamic adaptation of the training content for personalised trainings, by incorporating prior knowledge and context data into self-learning algorithms; (2) mapping of sensor data onto what happens in the VR environment, by focusing on motion prediction techniques that use past movements of the users, and (3) investigating immersive environments with intuitive interactions, by gaining a better understanding of human motion in order to improve interaction. The designed improvements will be characterised though a prototype VR training platform for multiple use cases. This work will not only advance the state of the art on VR training, but also on online e-learning applications in general.

Accurately predicting where the user of a Virtual Reality (VR) application will be looking at in the near future improves the perceive quality of services, such as adaptive tile-based streaming or personalized online training. However, because of the unpredictability and dissimilarity of user behavior it is still a big challenge. In this work, we propose to use reinforcement learning, in particular contextual bandits, to solve this problem. The proposed solution tackles the prediction in two stages: (1) detection of movement; (2) prediction of direction. In order to prove its potential for VR services, the method was deployed on an adaptive tile-based VR streaming testbed, for benchmarking against a 3D trajectory extrapolation approach. Our results showed a significant improvement in terms of prediction error compared to the benchmark. This reduced prediction error also resulted in an enhancement on the perceived video quality.