|
Abstract
|
Human elements are considered critical contributing factors to most casualties in the maritime sector (IMO, 2021). The Standards of Training, Certification, and Watchkeeping (STCW) Convention has been a major framework for promoting safety by addressing human elements that can potentially lead to incidents and accidents (IMO, 1978). Following the capsizing of the Herald of Free Enterprise in 1987, the IMO adopted International Safety Management (ISM) code through its resolution A.741 (18) under the Safety of Life at Sea (SOLAS). IMO further issued Resolution A.884(21) and Resolution A.1075(28) to prevent accidents related to seafarers' unsafe acts. In Resolution A.884(21), the IMO defines an unsafe act as “an error or violation that is committed in the presence of a hazard or potential unsafe condition” (IMO, 1999b). Decisions with no apparent resultant actions but having a negative impact on safety should also be considered as unsafe acts.
Despite continuous efforts from maritime regulatory administrations over the last few years, human elements still remain the most significant contributing factors to maritime accidents (Allianz, 2014; Chen et al., 2019; EMSA, 2017, 2020; Liu et al., 2021; Zhang et al., 2021). For example, according to the European Maritime Safety Authority (EMSA), over the 2014–2019 period, 54% of 1801 accident events were attributed to the ‘human action’ category compared to 28% due to ‘system/equipment failures (EMSA, 2020).
Therefore, understanding the occurrence and managing unsafe acts have been and continue to be essential aspects of the maritime sector. Research related to unsafe acts has focused on exploring the correlation between crew's unsafe acts and induced accidents (Fu et al., 2021) and error-producing conditions for unsafe acts (Kandemir and Celik, 2021). However, only a few studies have been conducted on early warnings of unsafe acts through detection and prediction.
Traditionally, the observation and identification of unsafe acts are performed manually. In recent years, advancements in computer vision-based methods and deep learning techniques have allowed high-risk industries to improve onsite safety performance by implementing automatic safety behaviour monitoring (Fang et al., 2020). However, the maritime sector has not fully utilised such technologies, even though experience reveals that timely reminders can significantly reduce the number of unsafe acts committed by the crew during voyages (Xu and Wang 2020). Human error prediction, detection, and timely correction play essential roles in preventing accidents (Reason, 2008). The bridge is a confined area with limited space available for installing equipment for behaviour-based monitoring. Meanwhile, several unsafe acts are difficult to recognize owing to technological constraints (e.g., unsafe speed, failure to follow the rules in sight of one another, and over-reliance on navigational equipment). Therefore, the main objective of this study is to explore a framework for developing a hybrid early warning system to either detect or predict unsafe acts during navigation....
Language: en |