A Research on Autonomous Collision Avoidance under the Constraint of COLREGs
Round 1
Reviewer 1 Report
This is another study on collision avoidance of autonomous ships. It builds on previous studies and adds a new algorithm and, most beneficial, knowledge about maritime reality and seamanship. As such it contributes to the knowledge building in this domain. Unfortunately, the article has been handed in prematurely and contains a number of unnecessary errors. I have in my comments also urged the author to be more candid about the problems of translating COLREGS qualitative enumerations to quantitative data for the algorithm to act upon. For instance, COLREG’s Rule 16, Action by Give-way Vessel, says: “Every vessel which is directed to keep out of the way of another vessel shall, so far as possible, take early and substantial action to keep well clear.” The algorithm in this study handles this situation. But, how has the time, distance and rudder angle for “early” and “substantial” been decided?
Having said that I congratulate the author to an interesting study.
The comments below (together with other comments, not present here, are made in the attached commented version of the paper).
Page 1, line 14: What do you mean by “steering time”? Is it the time it takes to turn from one course to another? ("turn time"). Or is it the time from the wheel-over command until the ship is on the new course? Please clarify. (After having read your paper, I think you might mean: “start of evasive action”?)
Page 1, line 15-17 you write: “The simulation results show that the method [---] can consider the unpredictable strategies for other vessels.” I do not see that you have shown this in your study. What you have shown is that your algorithms can consider predictable strategies of other ships keeping course and speed throughout the scenarios shown. For instance, if the overtaking target ship in Figure 5 b, suddenly changes her mind and decides to overtake on the port side, how would your algorithm react? If you can show this, then you have covered "unpredictable strategies of other vessels. If you want to keep this claim in the abstract, you should address this in the text and in the conclusions.
Page 1, line 20: “1. Introduction”. "1" Section numbering in scientific papers generally starts with “1”, not “0”. The numbering of all other sections should change accordingly.
Page 2, line 73: You write “They simulated several situations for OS and target ships.” What does “They” refer to here? Also, “OS” should be explained the first time used. I would suggest writing “Several situations for “own ship” (OS) and “target ships” (TS) where simulated.”
Page 3, line 96: Figure numbering should start with “1”.
Page 3, line 103: Again, "steering time" is ambiguous.
Page 4, line 115: Brackets are missing around a reference. Should be "stage [21]".
Page 4, line 128: Table 2. with the parameters for the simulated container ship is missing.
Page 5, Figure 2: You should mark which track belongs to “OS” and which to “Target Ship” in the Figure.
Page 5, line 142: You re here for the first time mentioning “OS's domain”. You are using the concept of "ship domain", you also mention this several times later in the paper. You should introduce this concept properly (maybe here).
As I understand it you are using this concept to trigger evasive maneuverer and thereby quantifying the qualitative enumerations of COLREG's "in good time", “ample time” and "early action” (Rule 8) and "early and substantial" (Rule 16). Because you also claim your algorithm considers “good seamanship” I think it could be appropriate that you made some comments on the problem of translating COLREGs qualitative enumerations into quantitative numbers.
Page 6, lines 169 and on: Here you write about “diversity of perception modes between crewed and uncrewed ships”. This is an opinion piece should be kept in a separate section titled "Discussion". Precisely how this problem should be solved is presently subject for discussion in the IMO and elsewhere.
I think you should keep the figure (maybe adding which COLREG's rules you are referring to). But keep the comments to what is stated in COLREGS for Section II "Conduct of vessels in sight of one another" and Section III "Conduct of vessels in restricted visibility".
Page 8, line 190: In this section you several times refer to "Figure 2". I presume you mean Figure 3.
Page 8, line 191: You write “After forming a close quarters situation, OS can take corresponding measures, generally turning against the target ship.” I think you are using the wrong word here: It should be either "towards" or "away from". In this case I think you mean "away from". However, I would say that the qualifications of Rule 17, in (ii) and (b) is only applicable when the give-way ship is not taking appropriate action.
Page 8, line 201: “After the close quarters situation is formed, measures such as turn to starboard should be taken,…” I would suggest "may be taken" along with Rule 17 (ii): "The latter vessel may, however, take action to avoid collision by her manoeuvre alone, as soon as it becomes apparent to her that the vessel required to keep out of the way is not taking appropriate action in compliance with these Rules." Only 17 (b) uses the word "shall", but in a very specific case: "When, from any cause, the vessel required to keep her course and speed finds herself so close that collision cannot be avoided by the action of the give-way vessel alone, she shall take such action as will best aid to avoid collision."
Manoeuvring after 17 (b) all the time might be a hazard because then ships can skip taking evasive action, knowing that the autonomous ship will in any case give way. The general rule is "keep course and speed".
In any case you should be open about how and why you have made your design decisions.
Page 8, line 223: You talk about “turning time”. Again, do you mean the time to commence an evasive manoeuvre? "Turning time" is the time it takes to make a turn?
Page 8, line 227: You write “When the time goes TCPA, the…” This is not correct English and is not understandable. Please, rephrase.
Page 9, Figure 5: To make the screen shots more understandable you should (a) mark which track belongs to OS and target ship, (b) add an arrow so it is clear in which direction the ships are traveling and (c) add a couple of time markers (t1, t2, etc.) on each track (and in Figure 6) so that the position of each ship at each time step becomes clear. It would also make the distance curves in Figure 6 useful.
Page 9, line 242-255: Please enrich the text and comment on how you use the “ship domain” ellipses visible in Figure 5 a-d.
Page 10, Figure 6: The curves in this Figure can be made more useful if some selected time markers where added that connected some instances in these curves with positions on the tracks in Figure 5.
Page 10, line 278: You write: “If not, increase the range of collision avoidance measures.” What do you mean by this? Increase the turn angle? Or add new measures (like slowing down or stopping)?
Page 14, Reference section there are some “[J]”s and [C]”s after the title of papers in the reference list (Journal and Conference?). They may be your personal comments, but should not appear in the reference list.
Comments for author File: 
Comments.pdf
Author Response
Dear reviewer
Thank you very much for your review. Your comments have been fully considered and adopted.
best ragards.
Author Response File: Author Response.pdf
Reviewer 2 Report
I found this paper very interesting and pretty well written (it does need another pass for some minor grammar and typographical errors). I like the premise and the research done.
I do question the necessity of the early sections detailing the different mathematical models. While interesting, it was unclear how the model directly impacted the rest of the paper. The meat of the paper seemed to start with Section 2.
Another reference worth considering related to autonomous ships and COLREGS is:
Meyers, J. (2022, Spring). COLREGS and Autonomous Surface Vessels. Proceedings of the USCG Marine Safety & Security Council, 79(1), 46-52. https://www.dco.uscg.mil/portals/9/DCO%20Documents/Proceedings%20Magazine/Archive/2022/Vol79_No1_Spr22.pdf?ver=dt8USTcoaX1ac8wzW7Q2ng%3d%3d
Author Response
Dear reviewer
Thank you very much for your review. Your comments have been fully considered and adopted.
best ragards.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper is tackling a fundamental topic: how the autonomous shipping technologies can ensure the navigation safety in compliance with the existing regulations (COLREG issued by IMO), in conditions of autonomous and manned mixed shipping traffic. The paper is excellent: well organised, referenced and descriptive, methodologically robust, nonetheless easy to understand for the reader, despite the natural complexity of the problem and the solution proposed. Ony two suggestions:
- introduce the acronym OS from the very beginning;
- improve the quality of figures concerning the distance durations (6, 9 and 10), possibly linking them with the trajectories in fig. 5 and 8.
Author Response
Dear reviewer
Thank you very much for your review. Your comments have been fully considered and adopted.
best ragards.
Author Response File: Author Response.pdf
Reviewer 4 Report
After review this manuscript, some revisions have been provided as following.
(1) In introduction, the literature review should be rewritten. Note, it is not the literatures listing.
(2) The language is too colloquial, such as in abstract, and in page 2 (2) and (3).
(3) the influence of hydrodynamic force on steering time during ship steering is not well represented in the algorithm.
(4) In the manuscript, the control algorithm is not provided.
(5) For the CA, many results have been obtained, so the difference of this paper should be detailed described.
(6) Figures 5,6,8,9, are the simulation result from the simulator, and they can be obtained easily. The other vital figures should be added, such as the control input, or the heading angle.
Author Response
Dear reviewer
Thank you very much for your review. Your comments have been fully considered and adopted.
best ragards.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Congratulations, version 2 has been much approved, in my opinion and I have just some minor comments that I would like you the consider before I recommend the paper to be accepted.
Some smaller language changes are only commented in the attached file.
A general question is if it is common practice at the DMU to talk about “DCPA” instead of just “CPA” (which would mean precisely the same: “distance to coarsest point of approach”) which in my experience is most used in maritime academies. (But I see that DCPA is common in mostly Chinese papers).
Page 5, Figure 2: There are some mismatches with the ASCII codes. On my computer, the apostrophe (') becomes another character (Ď). See the screen shot at the end of the attached PDF file.
Page 6, line 161: After “enters the OS’s domain” you could add "(see section 3.2)".
Page 7, Section 3.2 OS’s domain: Very good addition! But you could comment on why it is a circle in Figure 2 and here you write it is an ellipse? You could also add some comment on why you have chosen these particular numerical values. And maybe mention the connection to COLREGs qualitative enumerations "early", "in ample time", etc.
Page 7, line 205: When talking about “sight” you could clarify your arguments by mentioning "human vision" as opposed to “sensor vision”. But it is up to you. Otherwise the section is fine as it is.
Page 11, Figure 6: It is very good that you have added two examples of how your algorithm successfully handled unpredictable actions by target ships.
In your reply to my earlier comments you mentioned that you have added " (a) mark which track belongs to OS and target ship, (b) direction makers and (c) time markers". I only see OS and TS markers.
And the ship domain ellipses are now missing in (a)-(e)
Author Response
Response to Reviewer 1 Comments
Point 1: A general question is if it is common practice at the DMU to talk about “DCPA” instead of just “CPA” (which would mean precisely the same: “distance to coarsest point of approach”) which in my experience is most used in maritime academies. (But I see that DCPA is common in mostly Chinese papers).
Response 1: I strongly agree with your opinion. In our university, we usually use ”DCPA” to express “distance of the closet point of approach”. It has the same means of “CPA”
Point 2: Page 5, Figure 2: There are some mismatches with the ASCII codes. On my computer, the apostrophe (') becomes another character (Ď). See the screen shot at the end of the attached PDF file.
Response 2: The figure has been adjusted accordingly.
Point 3: Page 6, line 161: After “enters the OS’s domain” you could add "(see section 3.2)".
Response 3: I added "(see section 3.2)" in the sentence.
Point 4: Page 7, Section 3.2 OS’s domain: Very good addition! But you could comment on why it is a circle in Figure 2 and here you write it is an ellipse? You could also add some comment on why you have chosen these particular numerical values. And maybe mention the connection to COLREGs qualitative enumerations "early", "in ample time", etc.
Response 4: I added a sentence “In open waters, the ship domain can be set as a circle (see Figure. 2), while in restricted waters, the ship field is generally set as an ellipse. According to the mariner's habits in the coastal waters, …”.
Point 5: Page 7, line 205: When talking about “sight” you could clarify your arguments by mentioning "human vision" as opposed to “sensor vision”. But it is up to you. Otherwise the section is fine as it is.
Response 5: I added the sentence “Compared with human vision, sensor vision has a longer detection distance and higher accuracy. As a result,” at this point.
Point 6: Page 11, Figure 6: It is very good that you have added two examples of how your algorithm successfully handled unpredictable actions by target ships.
Response 6: Thanks a lot.
Point 7: In your reply to my earlier comments you mentioned that you have added " (a) mark which track belongs to OS and target ship, (b) direction makers and (c) time markers". I only see OS and TS markers. And the ship domain ellipses are now missing in (a)-(e)
Response 7: please see the figure below. The ship domain ellipses have been added in (a)-(e)
Thank you very much for your review. Your comments have been fully considered and adopted
Author Response File: Author Response.pdf
Reviewer 2 Report
A very minor point on terminology. On p. 2, lines 77-78, you write: "For example, in case of a head-on situation where OS turns to the starboard, overtake TS from the port side;"
Per the COLREGs, in a head-on situation, both vessels should turn to starboard. Since they are in a direct approach towards each other, neither is overtaking the other; they pass each other. Thus, a better wording might be "For example, in case of a head-on situation where OS turns to the starboard and passes TS on the port side;"
Author Response
Response to Reviewer 2 Comments
Point 1: A very minor point on terminology. On p. 2, lines 77-78, you write: "For example, in case of a head-on situation where OS turns to the starboard, overtake TS from the port side;"
Response 1: I adjusted the sentence ” For example, in case of a head-on situation where OS turns to the starboard and passes TS on the port side;”.
Thank you very much for your review. Your comments have been fully considered and adopted
Author Response File: Author Response.pdf
Reviewer 4 Report
I have no comments
Round 3
Reviewer 1 Report
The paper v.3 have now been considerably improved and I can gladly recommend publication in the present form.
I have attached the commented version of your last (v.2) manuscript, since it did not reach you. Most of my comments you have already answered, but there are some additional issues and clarifications which you may find valuable. You can consider them, or not, as you please.
Congratulations to an interesting study.
Comments for author File: Comments.pdf
