Computer-Human Mutual Training in a Virtual Laboratory Environment

Abstract

Science universities face the constant challenge of training their students in making appropriate use of their laboratories while avoiding accidents or equipment damages. Such a task becomes even harder for universities offering distance education, as their students visit less often their lab facilities and may have limited opportunities to become familiar with the respective instruments and equipment. For this purpose, the Hellenic Open University, which offers a long-standing distance education program in natural sciences, has been developing Onlabs, an interactive 3D virtual lab resembling its on-site biology laboratory for its students to train before they actually conduct live experiments. Recent versions of Onlabs contain, among others, an Instruction Mode, in which the human user is being guided by the computer to conduct a particular simulated experiment, and an Evaluation Mode, in which the computer evaluates the performance of the human user with respect to the completion of an experiment, contributing further to an effective learning process. Hence, in order for the performance assessment to be accurate, two separate machine learning techniques, a genetic algorithm and back-propagation on an artificial neural network, have been used.