Multi-robot Interaction with Mixed Reality for Enhanced Perception
Résumé
The growing development of Mixed Reality (MR) simulators in diverse areas such as vehicle operation, sports, and healthcare underscore the importance of high-quality immersion. Currently, evaluating the immersion quality of these simulators relies solely on post-simulation feedback from users, typically gathered through subjective questionnaires. However, this method fails to capture essential data for estimating the user's self-motion perception, which is crucial for optimizing immersion non-empirically. To address this gap, we propose a human homothetic perception framework. This framework aims to replicate human self-motion perception capabilities within a multimodal robotic system in MR environments. It incorporates a hybrid Gough-Stewart platform with a human-substitute NAO robot. This setup aims to recreate motion perception by integrating visual and vestibular information in a human perception model. We demonstrate this framework using trajectories from a Unity-based virtual ski simulator, designed to rehabilitate individuals with disabilities. The quality of immersion is evaluated by comparing the motion perception errors of the human model against the virtual trajectories. This innovative approach offers the first calibration tool for MR applications, eliminating the need for human feedback.