Costs and Beneﬁts of Autonomous Shipping—A Literature Review

: The development of autonomous ship technology is currently in focus worldwide and the literature on this topic is growing. However, an in-depth cost and beneﬁt estimation of such endeavours is in its infancy. With this systematic literature review, we present the state-of-the-art system regarding costs and beneﬁts of the operation of prospective autonomous merchant ships with an objective for identifying contemporary research activities concerning an estimation of operating, voyage, and capital costs in prospective, autonomous shipping and vessel platooning. Additionally, the paper outlines research gaps and the need for more detailed business models for operating autonomous ships. Results reveal that valid ﬁnancial models of autonomous shipping are lacking and there is signiﬁcant uncertainty affecting the cost estimates, rendering only a reliable evaluation of speciﬁc case studies. The ﬁndings of this paper may be found relevant not only by academia, but also organisations considering to undertake a challenge of implementing Maritime Autonomous Surface Ships in their operations.


Introduction
Recently, we can observe a continuously growing trend worldwide in research and development of autonomous ships in academia, industry, and maritime authorities. As per International Maritime Organization [1], autonomous ships are referred to as Maritime Autonomous Surface Ships (MASS). They are expected to operate in one of four degrees of autonomy, dynamically switching between them, as presented in Table 1. The classification was prepared only for a regulatory scoping exercise aimed at identifying legal barriers to MASS implementation, but is also used for other purposes [2,3].Currently, the autonomous ships are not allowed for international shipping due to law regulations [4][5][6][7].

Ship with automated processes and decision support
Seafarers are on board to operate and control shipboard systems and functions. Some operations may be automated.

Remotely controlled ship with seafarers on board
The ship is controlled and operated from another location, but seafarers are on board.

Remotely controlled ship without seafarers on board
The ship is controlled and operated from another location. There are no seafarers on board.

Fully autonomous ship
The operating system of the ship is able to make decisions and determine actions by itself.
stemming from implementing these solutions, compared to other aspects, for example safety [24][25][26][27][28][29][30][31][32][33][34][35][36][37][38]. The descriptions of economic attractiveness of the projects described earlier in this section do not contain the detailed information about components affecting the economic viability of the anticipated developments. These encompass the following: insurance issues, the costs of operating the ocean-going autonomous containerships, the cost of maintenance and repairs on board the ship during voyage, the increasing cost of protection against cyber piracy, monetization of NO x and CO 2 emissions, and the size of an autonomous fleet profitable for the shipowner. Therefore, the aim of the study is to determine the cost categories related to operation of autonomous ships (CAPEX and OPEX), including the costs of a Shore Control Centre allowing for safe navigation. To this end, a literature survey is performed to establish the state-of-the-art system concerning the use of autonomous shipping in the economic context.
The paper is structured as follows. In Section 2, the methods used in the study are described. Section 3 presents the results, including state-of-the-art qualities in economic issues concerning bulk and container vessels. A discussion is provided in Section 4, while Section 5 summarizes and concludes the paper.

Materials and Methods
This section presents the framework adopted to conduct our study and the analysed materials. In order to collect the latter, a three-stage procedure was adopted.
A systematic approach was applied for data gathering. Second, the data sample was filtered (preliminary and finally). The first stage of collecting the database assumed introducing an appropriate search query. Preparing a dataset was initiated by selecting a proper database. The articles were gathered from Google Scholar, which is widely acknowledged as a freely accessible and inclusive database of scholarly literature. Papers were gathered with the use of the following query, where the minus sign denotes the exclusion of a given word from the search. economy of autonomous vessel, business model, cost, unmanned vessel, -aerial, -underwater, -rov, -drone, -vehicle, -UAV, -seismic, -medicine, -robot, -history, -cyber, -bitcoin, -oceanographic, -combat, -navy, -offshore, -spacecraft, -airspace, -fish The query was conducted on 1 October 2020, and there was no limit for the year of publication in the search. As a result, an initial database was developed, consisting of 193 scientific documents.
The second stage of preparing the data sample was investigating titles and keywords concerning the papers gathered as the initial database. After browsing titles and abstracts, 174 documents were rejected since they did not refer to MASS, thus, leaving 19 documents for the final stage. A 90% rejection ratio is not atypical for reviews based on manual browsing [8].
In the third stage, the 19 papers were reviewed in depth. In this procedure, 15 papers presenting costs and economic benefits of autonomous shipping were selected, while four were dismissed, due to no significant reference to the analysed subject. Moreover, during this final stage of filtering, attention was also given to the lists of references of the finally filtered articles, since these could provide additional papers, which were not found at earlier stages. After browsing the titles, some of the articles were considered as relevant to the analysed subject. As a result, four additional articles were included in the final data sample, yielding 19 documents for the analysis. The entire process of data gathering is depicted in detail in Figure 1. The list of documents gathered during second filtering is presented in the Appendix A. earlier stages. After browsing the titles, some of the articles were considered as relevant to the analysed subject. As a result, four additional articles were included in the final data sample, yielding 19 documents for the analysis. The entire process of data gathering is depicted in detail in Figure 1. The list of documents gathered during second filtering is presented in the Appendix A.
The breakdown of the analysed documents according to publication year and type of publication is presented in Table 2.

Year
Number The breakdown of the analysed documents according to publication year and type of publication is presented in Table 2.

Literature Overview
The analysed literature shows that numerous documents raise the issue of costs and the effects of implementing autonomous shipping in the economic aspect. It is considered relevant for the domain by many authors. However, only 6 out 19 analysed publications tackle this issue in a structured and detailed manner. The in-depth description of costs in autonomous shipping is presented in Table 3. The remaining 13 articles, which are not listed in Table 3, contain only a brief and superficial information about the cost of autonomous shipping without indicating cost components. However, only two papers deal directly with the costs' estimation for autonomous ship operations, as presented below. One document presents literature analysis on selected aspects of autonomous shipping in terms of design, safety, navigation control, and short description of economic issues [39]. The rest of the papers focus on the Liner Shipping Network Design concept [23,40] and vessel platooning concept [41].
In the work by Lutz Kretschmann, Burmeister, and Jahn 2017, the operating, voyage, and capital costs were estimated for conventional bulker and her autonomous counterpart of the same capacity. The authors assume that maintenance of the ship will be conducted by crews in ports. Reduction of costs regarding crew wages and other related costs is compensated by the cost of running the Shore Control Centre, which is estimated to monitor 90 vessels at once. In their study, authors considered three scenarios: (1) reduced crew, (2) reduced crew and increased fuel efficiency, and (3) reduced crew, increased fuel efficiency, and the use of high-grade fuel. The cost of emergency arrival of the crew on board is not taken into account.
The case of an autonomous container ship for short sea trades with the capacity of 600 TEU was also considered. The total distance per round voyage was 1937 nautical miles [42]. The authors introduce communication costs that are estimated to increase ten-fold in comparison to a conventional container ship (see Table 4). The analysed literature focuses on presenting costs of construction and operation of autonomous ships as a selected case study of a container ship or bulk carrier. These are presented in Section 3.2. Estimating costs at the level of general assumptions is not performed in the analysed documents.

Basic Costs of Operating Autonomous Ships
The following section provides a detailed description of the economic performance of autonomous ships in two selected case studies. The analysis is focused on the cost of operating a bulk carrier and a containership.

Bulk Carrier
The basic motivation for the MUNIN project was to contribute to the triple bottom line-economic, social, and environmental issues and evaluate the project performance with the use of those three factors. The major challenges of the maritime industry according to Burmeister are: keeping expenses at the possibly low level, reducing greenhouse gas emissions caused by shipping, and removing crew from trivial ship operations [34].
The operation of the autonomous bulker with regard to MUNIN assumes fitting the ship with an Autonomous Navigation System, while the ship is controlled by the Advanced Sensor Module and supervision over the course of the ship's route is provided by the Shore Control Centre [21]. In a situation where Autonomous Ship Control is unable to cope with the difficulties encountered, the ship is switched to a remote control mode and the operator in the Shore Control Centre takes control over the ship [43]. First, the MUNIN concept assumed participation of the crew when departing and entering the port. However, the cost-benefit analysis alone indicated that the crew can feasibly be replaced by the use of a remote control for operations in port. Clearly, the question on safety of such a solution remains open [44]. The lack of crew on board will affect the cost structure. Consequently, the need for ship facilities intended for seafarers will not be necessary.
Moreover, the cost for equipment, annual expenses on rent, and operational costs of the autonomous bulker are given as rough estimates. While the investment cost of developing Shore Control Centre ranges from 1 to 2.1 mUSD, and the annual operating cost yields 0.87 mUSD [21,42].
The division of costs into operating, voyage, and capital of autonomous ship including reduction and increases with regard to implementing autonomous bulk carrier are presented in Tables 5 and 6, as per [21]. Table 5 shows the estimates of costs of operating the unmanned bulker, as per [21]. These are compared to a reference vessel-conventional bulker (length-230 m, breadth-32 m, design draught-14.5 m, service speed-15.5 kn, main engine-10.230 kW).  Table 6. Costs of running autonomous bulker-based on [21].

Operating Costs Voyage Costs Capital Costs
Crew Therein, the largest part of total costs (53%) relates to voyage costs (mainly fuel consumption), followed by the capital costs (26%), understood as autonomous technology and redundancies, while the operating costs accounts for 20% of the total cost. The authors claim the voyage costs are governed by two major factors: fuel consumption (up to 77%) and port-call (20%). The former can be reduced in case of the autonomous ship by approximately 6% owing to the reduced light ship weight, air resistance, and absence of the hotel system [45]. The required propulsion power could be reduced by approximately 1% if deckhouses were removed. Additionally, changes in light ship weight, caused mainly by the absence of extensive accommodation facilities, contribute to the reduced fuel consumption. The voyage costs also include boarding of a berthing party, which is estimated to be 20% higher than in case of a conventional vessel.
The lack of crew on board the autonomous ship, on one hand, results in the reduction of crew wages and crew-related costs (e.g., safety equipment), but, on another, it introduces a necessity of development and operation of the Shore Control Centre, which is estimated as 33,000 USD per vessel per year. All the repairs are expected to be carried out in ports by the specialized crew, which generates additional costs.
However, in Table 6, a qualitative assessment of relative changes in the given group of costs is provided for an autonomous ship, indicating which type of costs are expected to increase and what costs are expected to decrease, compared to a conventional bulk carrier.

Container Vessel
Currently, investment projects dedicated to autonomous container short-sea shipping are subjects of growing interest, mainly due to the continuously growing volume of container transport [46]. One of the projects related to autonomous container ships is the ReVolt, where a concept vessel designed by DNV GL along with its model have been developed [17]. Additionally, individual case studies are present in the literature [42].
The reduction of crew, thus, the living quarters, is expected to lower the capital costs of a small container ship feeder by 5% [47], which will also enable the ship to carry more containers on-board, as the ship is expected to gain up to 20% container slots [17]. Moreover, the removal of the superstructure may lead to lower port fees and, hence, decreased voyage costs [42]. Moreover, there are various modes of ship operation affecting the anticipated costs and benefits. One reads the ship will operate with the speed of 6 knots and a range of 100 nautical miles, while another claims the speed of 16 knots and an operational range of 2000 nm. In addition, the carrying capacity ranges from 100 TEU to 600 TEU. It is claimed that the autonomous feeder vessel compared to the conventional one generates savings USD 34 million during 30 years of operation. Thus, it is implied that the economic viability ceases to exist after that time of service [17]. However, the studies of 600 TEU autonomous container feeder reveal that this period is shorter by 10 years [42]. The exact figures could only be determined once actual economic data is available, especially when taking into account the costs induced and the technological development.
Among the operating costs, the one burdened with significant uncertainty are related to the insurance, as the risk level of autonomous ships remain largely unknown to date. This, in turn, may hamper the development of autonomous ships [5,48]. The level of insurance fees may be dictated by the evidence regarding reduced risk [17].
The introduction of autonomous ships in navigation implies increased use of sensors. DNV-GL claims that the average number of sensors will be between 15,000 to 20,000 onboard the ship to monitor the safety-critical and other relevant systems [17]. The average cost of the sensor is estimated as USD 0.38 (compared to 2004, the cost was estimated at USD 1.30 level) [49]. In addition, the redundancies on the ship will be a necessity [50]. To ensure safe and efficient operations, it is postulated to at least duplicate safety-critical systems including communication and navigation, as well as power supply [9,51,52]. This move will lead to the higher costs of investments and operations [53].
Recently, it can be observed that the attractiveness of seafaring is declining and crewing can become more cost-consuming [54]. In case of autonomous ships, the Shore Control Centre is expected to require the personnel of two certified officers per shift [42]. The extent of increase in communication costs remains unclear. However, it varies from a 10% increase- [47]-up to a tenfold raise in comparison to conventional vessels (26k USD/year/ship) [42]. Clearly, this is another significant source of uncertainty.
The summary of factors discussed for the autonomous bulk carriers (Section 3.2.1.) and container vessels (Section 3.2.2.) is given in Figure 2.
The extent of increase in communication costs remains unclear. However, it varies fro 10% increase- [47]-up to a tenfold raise in comparison to conventional vessels ( USD/year/ship) [42]. Clearly, this is another significant source of uncertainty.
The summary of factors discussed for the autonomous bulk carriers (Section 3.2 and container vessels (3.2.2.) is given in Figure 2.
However, the estimates may change dramatically, depending on the operation m the given autonomous ship is involved in, especially if a liner shipping network des concept is applied as discussed in Section 3.3. [21,42]. However, the estimates may change dramatically, depending on the operation mode the given autonomous ship is involved in, especially if a liner shipping network design concept is applied as discussed in Section 3.3.

Costs Induced Due to Various Operation Modes
One of the analysed operational modes suitable for autonomous ships is called a liner shipping network design (LSND-A) in short sea shipping [23]. The objective of such a mode is to minimize the total cost of operating the fleet of ships, leaving operational decisions out of the scope. The liner shipping network design is aimed at designing a set of weekly services within a specific time constraint based on the demands-time limit, demand, and destination [55].
In this mode, a mother vessel and a set of autonomous daughter vessels are anticipated. In the case study presented by Akbar et al. 2020, a mother ship serves the hub ports whereas daughter ships serve small ports along Norway. The mother route starts in the continental hub port, Rotterdam, and serves the Norwegian main ports. The daughter routes are routes between Norwegian transhipment port and serve other Norwegian main and small ports. Akbar [23] prepared an analysis on the type and number of autonomous daughter ships necessary to meet the transport demand. The fleet of daughter ships is heterogeneous in terms of capacity, which allows adapting the autonomous ship to transport needs. Introduction of autonomous daughter ships reduce operational cost by 11%. The majority of savings is generated by reduced time charter cost (as a result of the absence of crew).
Nevertheless, the construction cost of a fully autonomous ship is estimated to increase by approx. 5%. As a result, the reduction in operational costs is between 9 and 13% [40].
Nevertheless, Akbar et al. (2020) considered the case of introducing the autonomous mother vessel into operation in LSND-A. The reduction of cost of operating the fleet is 20% in comparison to fully conventional model. The fuel cost is decreased by 10% [23]. In the scenario with the use of a conventional mother vessel and autonomous daughter vessels calling 22 ports, the authors assume that the introduction of autonomous daughter ships is expected to reduce total operational costs by approximately 11%. The majority of the savings (94%) is a result of reduced time charter costs (the crew cost is removed) and fuel costs (6%). The cost of cargo handling in the port is increased by 22% in case of autonomous ships in comparison to conventional vessels.
The fleet of autonomous daughter ships is larger, but the ships have a smaller capacity in comparison to the conventional fleet, which reduces overall costs [23].
Similarly to LSND, the concept of the vessel train denotes a manned master-ship and automatically controlled the following vessels [56,57]. In a semi-autonomous vessel platooning concept, the main cost saving is reduction in crew cost. The increased number of the autonomous following vessels (FV) influences reduction in the cost associated with individual FV. The economic feasibility is due to the length of the vessel train. The cost savings of operating autonomous FVs should be equal to or larger than vessel train dues. The concept is relevant for routes that have constant cargo flow and high demand for transport services [58].

Sensitivity Analysis of a Life-Time Cost Assessment
One of the most popular methods to assess the economic performance of the project is net present value. The indicator presents a difference of future cash flows (both inflow and out-flow) with a given discount rate and the initial capital investment [59]. Kretschmann [21] also presents the required freight rate (RFR) as an indicator of economic effectiveness. RFR is the aggregated cost of owning and operating a vessel divided by total cargo tonnage over a specific period of time [60]. RFR is a freight rate, which produces a zero net present value over the assumed life of the ship [61]. The lower the RFR for the autonomous ship is, the more cost-efficient the ship is in comparison to the conventional one.
Kretschmann, Burmeister, and Jahn (2017) analyses three different scenarios and compares the indicators with the conventional vessel over 25 years. The first scenario (Scenario A) concerns the aspect of a reduced crew [39]. The expected present value (EPV) is the total cost of owning and operating the ship. The expected present value of the cost of owning and operating the autonomous bulker is 0.5 mUSD lower in comparison to a conventional reference vessel. The required freight rate for the autonomous bulk carrier is 0.4% lower than the RFR of the conventional bulker, which indicates that the autonomous bulk carrier is favourable in economic terms.
The second scenario (Scenario B), also described as the base scenario, takes into account the aspect of reduced fuel consumption and reduced crew. In this scenario, EPV (Expected Net Present Value) for the autonomous bulker is 4.3 mUSD lower in comparison to a conventional one. The RFR of an autonomous ship (producing 0 NPV-Net Present Value) is 3.4% lower than the RFR of a reference ship.
The third scenario (Scenario C) is based on three elements: reduced crew, reduced fuel consumption, and the use of high-grade fuel. The EPV for autonomous ship is estimated as 19.2 mUSD higher than the EPV of the reference vessel, mostly as a result of higher fuel consumption costs. Consequently, the RFR of the unmanned vessel is roughly 15% higher than for a conventional vessel [21].
The results of sensitivity analysis carried out for the three analysed scenarios (Table 7) demonstrates that autonomous ships can be expected to become cost-effective, especially when improvements in ship's fuel efficiency can be achieved, in addition to reduction of crew costs, as presented in Scenario 3 above. As depicted in Figure 3, the benefits in such a case are higher for long voyages and in a high fuel price scenario. Scenario B depicted in Figure 3 considers the reduced fuel consumption and the absence of crew during a sea passage. On the other hand, if the use of HFO (Heavy Fuel Oil) as the main fuel for an autonomous ship would not be possible. Then it would be extremely challenging for the vessel to be cost-competitive, unless she would be operated exclusively in SECA zones (Sulphur Emission Control Area). (Scenario A) concerns the aspect of a reduced crew [39]. The expected present value is the total cost of owning and operating the ship. The expected present value of th of owning and operating the autonomous bulker is 0.5 mUSD lower in compariso conventional reference vessel. The required freight rate for the autonomous bulk c is 0.4% lower than the RFR of the conventional bulker, which indicates tha autonomous bulk carrier is favourable in economic terms. The second scenario (Scenario B), also described as the base scenario, take account the aspect of reduced fuel consumption and reduced crew. In this scenario (Expected Net Present Value) for the autonomous bulker is 4.3 mUSD low comparison to a conventional one. The RFR of an autonomous ship (producing 0 N Net Present Value) is 3.4% lower than the RFR of a reference ship.
The third scenario (Scenario C) is based on three elements: reduced crew, re fuel consumption, and the use of high-grade fuel. The EPV for autonomous s estimated as 19.2 mUSD higher than the EPV of the reference vessel, mostly as a re higher fuel consumption costs. Consequently, the RFR of the unmanned vessel is ro 15% higher than for a conventional vessel [21].
The results of sensitivity analysis carried out for the three analysed scenarios 7) demonstrates that autonomous ships can be expected to become cost-effe especially when improvements in ship's fuel efficiency can be achieved, in addit reduction of crew costs, as presented in Scenario 3 above. As depicted in Figure  benefits in such a case are higher for long voyages and in a high fuel price sce Scenario B depicted in Figure 3 considers the reduced fuel consumption and the ab of crew during a sea passage. On the other hand, if the use of HFO (Heavy Fuel Oil) main fuel for an autonomous ship would not be possible. Then it would be extr challenging for the vessel to be cost-competitive, unless she would be operated exclu in SECA zones (Sulphur Emission Control Area). When it comes to various modes of autonomous ship operation, the liner shi network design concept demonstrated that its introduction is economically favou [23,40]. The authors also presented the sensitivity analysis for the case of a conven mother vessel and autonomous daughter vessels. As a result, [23] identified exceeding 119% of the additional construction cost of the autonomous daughte makes the operation of the autonomous ship economically infeasible. The sens analysis performed indicates that introducing autonomous daughter vessels is prof The authors focused on the construction cost of an autonomous ship in the case of 2 42 ports under normal demand. The introduced boundary of 119% is definitely h than existing cost estimates [23]. When it comes to various modes of autonomous ship operation, the liner shipping network design concept demonstrated that its introduction is economically favourable, [23,40]. The authors also presented the sensitivity analysis for the case of a conventional mother vessel and autonomous daughter vessels. As a result, [23] identified that exceeding 119% of the additional construction cost of the autonomous daughter ship makes the operation of the autonomous ship economically infeasible. The sensitivity analysis performed indicates that introducing autonomous daughter vessels is profitable. The authors focused on the construction cost of an autonomous ship in the case of 22 and 42 ports under normal demand. The introduced boundary of 119% is definitely higher than existing cost estimates [23].
Compared to a conventional container feeder, her autonomous counterpart is expected to face an internal rate of return at the level of 13.5% (the same as per conventional vessel), as presented by [42]. This is despite the newbuilding price for the autonomous ship being 32% higher than the conventional one (16.5 mUSD excluding the cost of SCC).

Discussion
The conducted literature analysis was based on the use of one database-Google Scholar. The query used in the study was formulated generally, with the simultaneous attempt to exclude issues related to other scientific disciplines. However, there were many articles in the dataset concerning both medical and social sciences, which is due to the polysemy phenomenon. Moreover, the limitation of the study was the utilization of different words to describe the same issues by authors of the articles than those used in the query. Due to the above-mentioned challenges, some of the relevant articles could have been omitted.
The herein performed literature review of the economic aspects of autonomous ships presents that much effort has been put into defining costs associated with the construction and operations of prospective autonomous ships, as well as developing relevant economic models for the aspects of the operation of the single ships as well as fleet of autonomous ships. However, there remain uncertainties regarding the costs, resulting mainly from the immaturity of the technology and various concepts of prospective ship and fleet operations. However, only after the introduction of autonomous ships into the shipping industry CAPEX and OPEX parameters can be determined.
The business models play a significant role in evaluating innovative transport projects. It is generally acknowledged that autonomous ships can bring about direct economic advantages. However, apart from direct gains, there are also non-trivial ones. It shall be noted that the propulsion of the ship in most cases is said to be environmentally-friendly. However, the benefits from the reduced emissions of NO x and CO 2 are not monetized within the existing models, while these issues gain ever-increasing attention [62][63][64]. Similarly, the costs related to safeguarding the cyber security [65,66] remains an open question, as well as contingency operations, or any other safety critical area of conventional shipping that presently requires direct intervention of humans in case of failure. All these directly affect the costs of insurance of prospective autonomous ships, about which the authors of the reviewed documents remained silent.
Another aspect is the economy of the scale. Meanwhile, the size of the fleet from the shipowner's point of view as the most favourable in cost-efficiency terms was not estimated for cases other than vessel platooning. Neither was the issue of how insurance rates would be affected by potentially different risk levels of autonomous ship operations [50]. Meanwhile, these account for as much as 1% of total operational costs of traditional vessels [67].
Taking into account the data presented in the previously mentioned documents, we can assume that preparing a detailed economic model concerning the cost breakdown related to the operation of the autonomous vessel fleet would allow for a determination of the directions of further development of autonomous shipping aimed at reaping from the implemented solutions. Preparing a general business model may become challenging regarding the differences in characteristics and market conditions of the areas where autonomous shipping could be implemented. It would be essential to focus on case studies concerning economic feasibility for introducing the specific autonomous shipping line or shipping connection [68].
The costs of potential accidents that the ship may be involved in is not accounted for. However, it is claimed that the prospective autonomous ships may be less likely accident-prone than the conventional ones [50].

Conclusions
The aim of the paper was to introduce the categories of costs that are essential when preparing economic analysis regarding autonomous ships based on the existing literature. To this end, the literature survey has been conducted, based on the Google Scholar database, covering 19 relevant documents published between 2014-2020.
As a result, the cost breakdown (operating, voyage, and capital) was prepared and the anticipated changes in the cost structure for both autonomous and conventional ships (containership and bulker) were presented. Additionally, the economic effect of various modes of ship operations, including vessel platooning, is discussed. Several gaps in knowledge became evident. The remaining open issues are the following: the cost of insurance, the costs of cyber security, or contingency operations. These are omitted in the literature most likely due to an absence of reliable data. Nevertheless, they should be considered in the analysis or any model of autonomous ship costs, since these may significantly influence the economic feasibility of such an endeavour.
The autonomous merchant ships are expected to serve one purpose above others-to haul cargoes between ports, bringing revenue to their owners and operators. Without an economic feasibility of the technology, its concept remains questionable from its very roots. To this end, the economical aspect of the technology shall be investigated, accounting for the existing uncertainties, in order to formulate clear suggestions as to the further direction of technological and operational advancements.
However, to perform such an analysis in a convincing and reliable manner, real-life data or alike are prerequisites along with an accurate description of the analysed system and its operational pattern as well as proper uncertainty treatment, communicating those to the end users. Even then, this will only be applicable to a particular case scenario, and generic assessment might prove extremely difficult to formulate due to the enormous complexity of the system itself.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1. List of 19 papers after second filtering.