Machine learning for human factors assessment in virtual environment

Abstract

Human factors are often associated with reducing error, enhancing safety, comfort and increasing productivity. Poor evaluation of human factors in working environments often result in human error which leads to the occurrence of critical accidents. In order to create an efficient, effective and safe working environment for workers and the development of appropriate prevention and mitigation strategies, studies conducted on fatigue levels and its influence on human error has been increasing over the years. Human factors also play a major role in individuals engaged in virtual environments, as the fatigue induced during the tasks carried out would affect the session and overall outcome. Hence, traditional questionnaires were introduced to gauge the levels of fatigue induced in the subjects when they are in a virtual environment. However, questionnaires only can be carried out pre and post experiment or session not giving space to any real-time monitoring of the participant’s fatigue levels for safety and future development reasons. This is where eye tracking data will help to give real time monitoring of fatigue levels. This will give us an easier, quick and efficient analysis of fatigue recognition in working individuals. This system will use raw data collected from Eye-tracking devices and process the data through machine-learning algorithms. Eye tracking however, consists of many different variables with some being reliable and some not. Hence, we will find out which variable in the raw eye tracking data has the highest level of accuracy, reliability and performance levels and also backed up by research for future use to get the most accurate fatigue recognition results.