As autonomous ships become more viable, appropriate risk indicators are increasingly needed. In the existing literature, few works are related to developing such risk indicators. Existing indicators are general and not focused on a specific autonomous vessel, let alone an actual operating ship. To bridge this research gap, this article proposes a methodology for identifying risk indicators based on a Bayesian belief network (BBN). The methodology is applied to hazardous event "losing navigational control" of a trial-operating autonomous passenger ferry. The risk indicators developed cover technical equipment, remote supervisors' capacity, and environmental conditions. The probability of losing navigational control is calculated considering the states of the risk indicators, which contributes to risk monitoring of the system during operation. Further, strong wind, the state of battery health, end-to-end delay, and packet-loss rate have been identified as critical risk indicators. These should preferably be presented in a shore control center to human operators and provide a basis for decision support, and for defining operational procedures for safe takeover and shared autonomy. The findings can further improve practitioners' understanding, monitoring, analysis, and management of the risk of loss of navigational control of autonomous ships, which is crucial to prevent collisions.
Discussion(0)
No comments yet. Be the first to comment.