Recognition of student intentions in a virtual reality training environment

This paper introduces a novel method for detecting and modeling intentions of students performing training tasks in a Virtual Reality (VR) environment enhanced with intelligent tutoring capabilities. Our VR-setup provides students with an immersive user interface, but produces noisy and low-level input, from which we need to recognize higher-level cognitive information about the student. The complexity of this task is amplified by the requirements of the target domain (child pedestrian safety), where students need to train complex skills in dynamic settings. We present an approach for this task, which combines the logic-based Event Calculus (EC) and probabilistic modeling.