Selfit – An Intelligent Tutoring System for Psychomotor Development

Abstract

Recent advancements in Machine Learning and software development are extending the applications of Intelligent Tutoring Systems (ITS) into non-cognitive skill domains, while proposing novel design architectures and tutoring strategies. In this paper we present our ongoing work on Selfit, an Intelligent Tutoring System for psychomotor development. The system focuses on and was tested for Anatomical Adaptation training, the first phase of training. The tutoring module includes a contextual multi-armed bandit algorithm for online generation of teaching sequences to overcome multiple problems, such as lack of training time, complexity of user characteristics, or management of motivation. First, the system was evaluated in a virtual environment, where populations of trainees follow several personalization strategies in systematic experiments. Second, Selfit is currently being tested by a group of users and a preliminary study revealed that most trainees find the system easy to use, modern, and attractive.