Machine Learning for Stress and Fatigue Detection in Ship Crews

The aim of this project will be to measure and predict the levels of fatigue and stress in targeted Deck Officers from modern LNG ships. This will be carried out using a range of both intrusive and non-intrusive methods. This will be combined with vessel and motions data and environmental data. Machine learning techniques (e.g., deep learning and Bayesian classifiers) can be used to determine periods of acute stress and poor sleep leading to fatigue. Based on these outputs, optimisation algorithms will be developed to reduce tiredness and stress levels. Moreover, this information can be used to develop training programmes incorporating appropriate stressors to reduce the negative effects of stress/ fatigue as the trainees become habituated. The project will also investigate the levels of activity and stress when off watch and when not at sea, in particular the periods before and after a long sea voyage.

The recent Global Marine Technology Trends 2030 report highlights increasing technology onboard ships and the need for highly skilled crew to operate them. The need to recruit highly skilled crew for ship operations will require significant development of training. The design of shipboard systems needs a multi-disciplinary approach building on work already carried out between engineering and psychology.

Working with our industrial partner, Shell Shipping and Maritime, this research has the potential to reduce major shipping accidents, saving lives and reducing environmental impact.