Machine Learning Application for Real-Time Simulator

This paper presents a groundbreaking research initiative that focuses on the development of an intelligent architecture for Adaptive Virtual Reality Systems (AVRS) in immersive virtual environments. The primary objective of this architecture is to enable real-time artificial intelligence training and adapt the virtual environment based on user states or external parameters. In a case study focused on detecting cybersickness, an undesired side effect in immersive virtual environments, we utilized this architecture to train an artificial intelligence model and personalize it for individual users in a driving simulator application. By leveraging the capabilities of this architecture, we can optimize virtual reality experiences for individual users, leading to increased comfort. We evaluated the system's performance in terms of memory usage, CPU and GPU usage, temperature monitoring, frame rate, and network performance, and our results demonstrated the efficiency of our proposed architecture.