Human-system concurrent task analysis for maritime autonomous surface ship operation and safety

https://doi.org/10.1016/j.ress.2019.106697Get rights and content

Highlights

  • Autonomous ships will rely on system-human interaction in the near future.

  • The human-system interaction in autonomy (H-SIA) method is proposed.

  • H-SIA consists of an event sequence diagram and a concurrent task analysis (CoTA).

  • CoTA allows to identify dependent tasks and failure propagation between sub-systems.

  • H-SIA method analyses the system as whole, rather than each component separately.

Abstract

Maritime Autonomous Surface Ships (MASS) are the subject of a diversity of projects and some are in testing phase. MASS will probably include operators working in a shore control center (SCC), whose responsibilities may vary from supervision to remote control, according to Level of Autonomy (LoA) of the voyage. Moreover, MASS may operate with a dynamic LoA. The strong reliance on Human-Autonomous System collaboration and the dynamic LoA should be comprised on the analysis of MASS to ensure its safety; and are shortcomings of current methods. This paper presents the Human-System Interaction in Autonomy (H-SIA) method for MASS collision scenarios, and illustrates its application through a case study. H-SIA consists of an Event Sequence Diagram (ESD) and a concurrent task analysis (CoTA). The ESD models the scenario in a high level and consists of events related to all system's agents. The CoTA is a novel method to analyse complex systems. It comprises of Task Analysis of each agent, which are preformed concurrently, and uses specific rules for re-description. The H-SIA method analyses the system as whole, rather than focus on each component separately, allowing identification of dependent tasks between agents and visualization of propagation of failure between the agents’ tasks.

Introduction

Research and development projects on Maritime Autonomous Surface Ships (MASS) have faced increasing interest, and some are currently in a testing phase. For instance, Yara Birkeland, an autonomous and electric container vessel developed by Yara and Kongsberg, is expected to undergo the first operational tests at the start of 2019, and to conduct fully autonomous operations by 2020 [1]. DNV ReVolt, an unmanned shortsea vessel developed by DNV GL, is being tested in a 1:20 scale, in collaboration with the Norwegian University of Science and Technology (NTNU) [2]. In addition, NTNU is currently testing a 1:2 scaled autonomous passenger ferry, which is expected to run on full scale in 2020 [3]. Several research projects and forums also address MASS, such as the Advanced Autonomous Waterborne Applications Initiative (AAWA) [4]; the Norwegian Forum for Autonomous Ships (NFAS) [5], [6] and The Maritime Unmanned Navigation through Intelligence in Network (MUNIN) [7].

The growing number of projects related to MASS is due to the expected advantages it may bring, compared to traditional manned ships. The removal of the accommodation and deckhouse can save cost, weight and space; enable the ship to carry more cargo [4], create more flexible transport solutions [6]; provide better accessibility to potentially dangerous areas [8]; and lead to greener shipping. Moreover, MASS’ operation may be safer than traditional manned ships, since human error may be a contributing factor to many marine accidents [9]. Less crew or unmanned ships may also lead to fewer fatalities and injuries if an accident should occur.

Nevertheless, a fully autonomous vessel with no supervision and/or interference from humans is not expected to be a reality in the near future. Most of the projects regarding MASS include operators working in a shore control center (SCC), whose responsibilities may vary from supervision to remote control. MASS operation will thus rely on a human-autonomous system (H-AS) collaboration. Moreover, MASS may have a dynamic Level of Autonomy (LoA): the LoA may change in the same voyage depending on certain conditions. Hence, the operators’ tasks may change during a voyage: they may have to control the vessel remotely during parts of the voyage, e.g., when maneuvering in a busy congested harbor, and then change to monitoring the vessel when it moves to the open sea.

As humans will still be involved in the operation at some level, human error may still occur [10], [11], [12]. In addition to human error, MASS introduce new challenges to the maritime sector, such as increased cyber security threats, the possibility of losing communication with the control center; or the difficulty of performing maintenance during sea voyages [4], [10]. Hence, risk assessments of MASS operation are important [13]. They face two main challenges: (i) the strong reliance on H-AS collaboration during the operation, and (ii) the possibility of a dynamic LoA.

Few publications address topics related to hazards and risks associated with MASS operation. A recent review [13] of risk models aiming at AS and conventional ships revealed that current models do not sufficiently model the functions carried out by software based systems and that human operators are often treated superficially. Different operational modes of vessels are only covered to a limited extent. The current literature aiming at MASS also does not model and analyse the H-AS interaction as potential contributor to the risk of MASS operation, nor does it reflect the dynamic LoA of the operation. This paper intends to fill this gap, through the development of a method for analysing the operation of unmanned autonomous ships.

The contribution of this article is the Human-System Interaction in Autonomy (H-SIA) method for MASS, which consists of two elements; an Event Sequence Diagram (ESD) and a novel method called concurrent task analysis (CoTA):

  • (i)

    The ESD models the scenario in a high level and consists of events related to both humans working in a shore control centre (SCC) and the autonomous ship system. We provide a flowchart, in which questions about the design and LoA guide the analyst to build the ESD. The flowchart ensures traceability and reproducibility and is well suited for MASS designed for operations with either low or high autonomy, or for a dynamic LoA;

  • (ii)

    The CoTA is a new method introduced in this paper to analyse complex systems. It is developed from the ESD. It models the interactions between tasks performed by different agents (e.g., humans and autonomous ships). A CoTA comprises of a Hierarchical Task Analysis (HTA) or Tabular Task Analysis (TTA) of each agent, in which the tasks are re-described until basic tasks that relate to the interaction between the agents are determined. Moreover, the CoTA uses a cognitive model and extends it to the complete system. The CoTA has several purposes, such as failure events identification, failure propagation analysis, and procedures development.

The innovative aspect of the H-SIA method is that the system is analysed as whole, rather than focused on each component separately. In addition, it may be used to compare risks from different MASS designs, or different LoAs during the design phase of the ship, as well as during operation. Further, the CoTA makes it possible to identify dependent tasks between different agents and to visualize propagation of failure between the agents’ tasks. Since the CoTA uses the IDA model, it is possible to identify failures related to both humans and the AS.

The paper is organized as follows: Section 2 presents the state of the art for autonomous ships risk models and an overview of the background used for the development of the H-SIA method, namely: the ESD, Task Analysis and IDA model. Section 3 presents the H-SIA method, its elements, and its advantages. The method has been developed for collision scenarios, as collision is one of the main causes to ship losses [14]. An application to a potential scenario demonstrates systematically the H-SIA method for MASS, with its strengths and limitations in Section 4. Section 5 concludes the paper.

Section snippets

Autonomous ships and risk assessment

Autonomy can be defined as “a system's or sub-system's own ability of integrated sensing, perceiving, analysing, communicating, planning, decision-making, and acting, to achieve its goals as assigned by its human operator(s) through designed human-machine interface (HMI)” [43]. Although the term “autonomous ship” suggests, at a first glance, a concept in which a ship would have a system that is fully responsible for all aspects of its navigation and is independent of humans, autonomous ships

H-SIA method for mass collision scenarios

The H-SIA method, presented in this Section, is composed of two elements: (i) an ESD (Section 3.1), and (ii) a CoTA (Section 3.2). The method is specifically developed for collision scenarios between an autonomous ship and another vessel or object, but is nevertheless expected to have general applicability for autonomous systems.

Fig. 4 presents the three main steps in the H-SIA method, and Steps 2 and 3 are detailed in the following sub-sections. The first step is familiarization with the

Step 1 – familiarization with the system

The case study consists of the following scenario:

  • The autonomous ship is initially operating in constrained autonomy (Level 7 in Table 1): it is responsible for detection and collision avoidance planning and execution. Its operation is supervised by a crew working on a SCC;

    • If the AS fails at detecting the CC, the crew at the SCC can detect it through monitoring the operation;

  • The AS detects another ship as a CC and generates a plan. It sends a warning to the SCC about the existence of the CC and

Method and case-study discussion

The scenario of the case study is relatively complex, since the AS operates with a dynamic LoA that can, according to the situation, shift from low LoA (remote control) to high (constrained autonomy). The application of the flowchart for the ESD development ensures that the foreseeable high level events related to the AS and to the operators are covered. The events from the ESD are successfully translated into tasks and developed into the CoTA using the rules provided in Section 3.2.1. This

Conclusion

This paper presents the Human-System Interaction in Autonomy method for MASS, developed for collision scenarios. The method consists of an ESD and a concurrent task analysis (CoTA) – a multipurpose technique introduced in this method. H-SIA considers the human-autonomous ship collaboration, an important part of autonomous ships operation, as well as the dynamic LoA these operations may have. The H-SIA method allows for analysing the autonomous system as a whole during different phases of design

References (50)

  • H. Alfheim et al.

    Development of a dynamic positioning system for the revolt model ship Henrik Alfheim

    (2017)
  • NTNU, “Autoferry – Autonomous all-electric passenger ferries for urban water transport,” 2018. [Online]....
  • Rødseth Ø, Nordahl H. Definitions for autonomous merchant ships, Norwegian Forum for Autonomous Ships-. NFAS,...
  • Rødseth Ø. Definition of autonomy levels for merchant ships, Norwegian Forum for Autonomous Ships-. NFAS,...
  • MUNIN, “Research in maritime autonomous systems project results and technology potentials,”...
  • AGCS, “Safety and shipping review 2017,”...
  • A.M. Rothblum

    Human error and marine safety

  • Ø.J. Rødseth et al.

    A risk based approach to the design of unmanned ship control systems

  • M Ramos et al.

    Accounting for human failure in autonomous ships operations, Safety and Reliability – Safe Societies in a Changing World

    Proceedings of ESREL

    (2018)
  • MA Ramos et al.

    Collision avoidance on maritime autonomous surface ships: operators’ tasks and human failure events

    Safety Science

    (2019)
  • C.A. Thieme et al.

    Assessing ship risk model applicability to marine autonomous surface ships

    Ocean Eng

    (2018)
  • S. Li et al.

    An overview of maritime waterway quantitative risk assessment models

    Risk Anal

    (2012)
  • GJ Lim et al.

    Models and computational algorithms for maritime risk analysis: a review

    Annals of  Operations Research

    (2018)
  • Ø.J. Rødseth et al.

    Risk assessment for an unmanned merchant ship

    TransNav Int J Mar Navig Saf Sea Transp

    (2015)
  • Cited by (89)

    View all citing articles on Scopus
    View full text