SOFIA: Advancements in Olfactory Stimuli in Virtual Reality for Immersive Multisensory Experiences and Potential Applications in the Industry 5.0

This paper presents advancements from the SOFIA (Sensorial Olfactory Framework Immersive AI) project, focusing on integrating olfactory stimuli into virtual reality (VR). The study highlights two core developments: AI models for olfactory mask segmentation and a miniaturized device for real-time chemical stimulation. Using annotated 360° images, the AI models were evaluated, with MAFT achieving superior performance due to its mask-aware architecture and GPT-4o mini assisted in label alignment. These findings underscore the potential for synchronized olfactory experiences in immersive VR. In parallel, the research assessed operational parameters for the olfactory device, such as fluid flow, concentration, and activation cycles, identifying critical factors for ensuring effective odor dispersion. While current efforts prioritize chemical stimulation, the project envisions future advancements in neural stimulation to eliminate the need for chemical vaporization. By refining AI models and device parameters, SOFIA lays the groundwork for transformative applications in health, entertainment, and education, enhancing multisensory VR interactions.