Deep-learning based recommenders for the improved user navigation in VR

Virtual reality (VR) has become a popular choice for education, industrial simulation, entertainment and healthcare applications. User navigation is an essential propriety of virtual applications. However, novice audiences often face the difficulty of engaging with the virtual surrounding environment. This work presents a novel design of a deep learning-based navigation solution to improve the quality of user experience and the engagement with virtual content. We compare two navigation methods avatar-based and arrows-based guidance, both of which are driven by a recurrent neural network (RNN) model. We capture participants’ mobility and eye-gaze to compare the impact of different navigation affects on users’ engagement in VR applications.