Risk-based supervisory control for autonomous ship navigation

• Published in: Journal of marine science and technology 2023-09, Vol.28 (3), p.624-648
• Main Authors: Blindheim, Simon, Johansen, Tor Arne, Utne, Ingrid Bouwer
• Format: Article
• Language: English
• Subjects: Algorithms; Analysis; Automotive Engineering; Autonomous navigation; Coastal environments; Coastal zones; Complexity; Control centres; Control systems; Crew; Decision analysis; Decision making; Engineering; Engineering Design; Engineering Fluid Dynamics; Liability; Mechanical Engineering; Navigation; Offshore Engineering; Optimal control; Optimization; Original Article; Qualitative analysis; Remote control; Risk analysis; Risk assessment; Risk management; Supervisors; Supervisory control; Systems theory
• ISSN: 0948-4280; 1437-8213
• DOI: 10.1007/s00773-023-00945-6

This paper proposes a novel method to transform the results of qualitative risk analysis into a numeric optimal control problem for autonomous ship navigation. Today, making autonomous high-level decisions replacing a crew onboard is considered difficult, in some part due to the complexity of managing the operational risks involved. Although human supervisors, e.g., located in remote operating control centers are still needed for safety and liability reasons, there is a growing demand for complex decisions to be made by the onboard control system itself, both during normal operations and in emergencies. This paper suggests general principles for how the results from systems-theoretic process analysis (STPA) can be transformed into a quantitative and computationally tractable optimization problem, solved by a MPC-based decision-making algorithm for autonomous navigation. The proposed method is demonstrated and evaluated by simulating an autonomous ship navigating in a coastal environment. It is concluded that the proposed method may serve as a reasonable and valuable bridge between the realms of qualitative risk analysis and numerical optimal control for risk-aware autonomous control and decision-making.