Plenary Session
Risk-aware autonomous systems for safe and intelligent decision making

The technological advances in automation and autonomous systems enable new and challenging missions, processes, exploration and monitoring of areas on land and at sea. Higher autonomy in tediousoperations may lead to safer and more efficient performance, supporting the human operator withrespect to workload, supervision, and decision making. Advanced control systems introduce, however,new types of failures, due to increased system and mission complexity and unforeseen functionalinteractions. With reduced involvement by a human operator, early warnings and predictions ofpotential deviations are needed to enable system reconfiguration and provide the human operator withenough time to intervene when necessary.

Risk perception is an important part of human cognition which currently is not adequately implementedin the control and decision-making of autonomous systems. Supervisory risk control is a novel researcharea which couples risk assessment and modelling with control system design, i.e., bridging thedisciplines of risk science, human factors, engineering cybernetics, and artificial intelligence (AI). Riskspecialists need to be an integral part of the development of AI and control systems to improve safetyand trust in these systems.

In this lecture, I claim that autonomous systems able to analyse and evaluate risks comprehensivelywould lead to highly intelligent system behaviours, safe interactions in challenging environments,andpublic trust and acceptance. I address challenges related to intelligent risk analysis and shared control,and promising solutions researched and tested using both simulation and field experiments with realrobotic platforms. The focus is on marine systems due to the demanding challenges in that domain, andmy experience as a researcher in the Centre of Excellence on Autonomous Marine Operations andSystems (NTNU AMOS), but the results should have relevance across different application areas.

Dr. Ingrid Bouwer Utne is a Professor at Department of Marine Technology at the Norwegian University of Science and Technology (NTNU). Her main research area is risk assessment and modeling of marine and maritime systems. Utne started her career in the marine domain as a young officer onboard two Norwegian frigates, and among other things she sailed with NATO's Immediate Reaction Force. Later, she has worked in the research institute SINTEF, in the industry, and she has been a visiting Scholar at UC Berkeley where she was a member of the Deepwater Horizon Study Group (DHSG) at the Center for Catastrophic Risk Management. The DHSG served as advisor to the US Presidential Commission, authorities, and the public on issues related to the Macondo blowout. In recent years she has specifically focused her research on improving the safety and intelligence of autonomous systems, as part of interdisciplinary work in the Center of Excellence on Autonomous Marine Operations and Systems (NTNU AMOS).