Next Article in Journal
Research on the Depth Control Strategy of an Underwater Profiler Driven by a Mixture of Ocean Thermal Energy and Electric Energy
Next Article in Special Issue
Use of Hybrid Causal Logic Method for Preliminary Hazard Analysis of Maritime Autonomous Surface Ships
Previous Article in Journal
A Comprehensive Star Rating Approach for Cruise Ships Based on Interactive Group Decision Making with Personalized Individual Semantics
Previous Article in Special Issue
Research on Ship Trajectory Classification Based on a Deep Convolutional Neural Network
 
 
Article
Peer-Review Record

Ship Intention Prediction at Intersections Based on Vision and Bayesian Framework

J. Mar. Sci. Eng. 2022, 10(5), 639; https://doi.org/10.3390/jmse10050639
by Qianqian Chen 1, Changshi Xiao 1,2,3,*, Yuanqiao Wen 2,4, Mengwei Tao 1 and Wenqiang Zhan 3
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
J. Mar. Sci. Eng. 2022, 10(5), 639; https://doi.org/10.3390/jmse10050639
Submission received: 15 March 2022 / Revised: 25 April 2022 / Accepted: 27 April 2022 / Published: 7 May 2022

Round 1

Reviewer 1 Report

The article deals with current and important issues that concern ship collisions. The article attempts to use data based on a Bayesian model to analyse maritime traffic problems, considering route distribution, traffic volume and regularity. The article has an adequate theoretical basis, relevant information and analysis, good partial (in the article) and final (in the conclusion) conclusions. The article uses original research by the author and cited research by other researchers, which enriches its content. It is written in good language and is based on an analysis of current and well-chosen literature. The introduction lacks a clearly stated thesis. The research models have been applied correctly. Systematics of models is not a simple issue, as the differentiation of model types results mainly from their precisely defined purpose. The article should therefore be regarded as an interesting introduction to a very important issue and treated as a scientific article.

Author Response

Answer: We appreciate the valuable comment by the reviewer. 

Reviewer 2 Report

The article deals with a very interesting issue that has great practical importance. Providing sufficiently early and reliable information about the intentions of maneuvering vessels in the vicinity will increase navigational safety. Although the article has a great potential to be published it requires both editorial and substantive changes.

Some general comments that should be taken into account to improve the quality of the article:

  1. The article is under-edited. Some examples: line numbering is missing in most of the article, the same sentence repeated on page 3, the caption of Figure 7 and its description, improper editing of part of equations, numerous typos (e.g., in additio, meta-automata, missing many punctuation marks and spaces).
  2. The paper lacks a technical description of the parameters, capabilities, and limitations related to the use of the proposed hardware solutions (especially radars).
  3. It is unclear how the proposed solution is going to be used, whether it is going to be installed and used on ships or rather enhance the capabilities of land-based traffic monitoring services.
  4. If the system is to be installed on the ship, the article should take into account the influence of the ship's own motion on the accuracy of predicting the maneuvers of other ships.
  5. The paper states that real-world experiments were conducted to evaluate the accuracy of the proposed model. The accuracy cannot be evaluated from the results presented because there is a lack of defined criteria for evaluating the accuracy of the model.
  6. It is written that presented method is “superior to the most advanced method”. In order to support such a statement, these methods should be presented, the criteria used for the comparison should be defined, and finally, the results should be presented.

 

Author Response

  1. The article is under-edited. Some examples: line numbering is missing in most of the article, the same sentence repeated on page 3, the caption of Figure 7 and its description, improper editing of part of equations, numerous typos (e.g., in addition, meta-automata, missing many punctuation marks and spaces).

Answer: We appreciate that the reviewer pointed this out. The manuscript has been proofread and improved carefully, including nouns’ single or plural, figure number, references, etc.

  1. The paper lacks a technical description of the parameters, capabilities, and limitations related to the use of the proposed hardware solutions (especially radars).

Answer: We appreciate that the reviewer pointed this out. The details about the technical description of the parameters, capabilities, and limitations related to the use of the proposed hardware solutions (especially radars) are described in Page 5 of the manuscript.

  1. It is unclear how the proposed solution is going to be used, whether it is going to be installed and used on ships or rather enhance the capabilities of land-based traffic monitoring services.

Answer: We appreciate that the reviewer pointed this out. At present, the scheme needs to input static environmental data of intersection waters, so it is only suitable for laying on the shore. Therefore, the supervision basis is added. In the future work, the problem of understanding dynamic navigation scenarios will be solved, and the static environmental data will no longer be input from the electronic chart, so that the scheme can be used on ships 

  1. If the system is to be installed on the ship, the article should take into account the influence of the ship's own motion on the accuracy of predicting the maneuvers of other ships.

Answer: Thanks. In future work, the influence of the ship's own motion on the accuracy of predicting the maneuvers of other ships will be considered when the scheme is used on ships.

  1. The paper states that real-world experiments were conducted to evaluate the accuracy of the proposed model. The accuracy cannot be evaluated from the results presented because there is a lack of defined criteria for evaluating the accuracy of the model.

Answer: We appreciate that the reviewer pointed this out. We will set evaluation criteria based on the accuracy and real-time performance of the predicted results, the details about the evaluation criteria are described in Page 15 of the manuscript.

  1. It is written that presented method is “superior to the most advanced method”. In order to support such a statement, these methods should be presented, the criteria used for the comparison should be defined, and finally, the results should be presented.

Answer: We appreciate that the reviewer pointed this out. As the most advanced existing methods for identification of ship intentions are mainly based on historical ship AIS data, data sampling frequency and integrity of the two methods can be compared to show that this method is more advantageous.

Author Response File: Author Response.pdf

Reviewer 3 Report

The work “Ship intention prediction at intersections based on vision and Bayesian framework” presents a methodology based on computer vision and radar data to identify the ship's navigation intention at intersections.

 

The work is interesting once it presents a viable alternative that can be easily implemented in many locations. However few improvements are required before publication.

 

The general comments are:

  1. First I would like to suggest an English review. For example:

A. “A dynamic Bayesian model is established which can accurately identify the ship's intention.”

The conjunction “which” should be replaced by “that”.

 

B. “Secondly, the RANSAC method is used to fit radar and image detection information, and the homographic matrix from radar coordinate system to image coordinate system is obtained.”

Here the article is missing “the radar”, “the image coordinate”.

 

C. “Although various technologies have been developed to prevent ship collisions, but according to a survey by the International Maritime Organization, more than 40% of shipwrecks worldwide are caused by ship collisions.” 

Seems that the phrase construction makes it strange with the “but” position. Using a direct order would be preferred.

 

D. “In additio, some studies inferred ship intentions by introducing factors such as static characteristics of the ship and the navigational environment to influence navigation.”

 

There is a missing “n” in the word addition.

 

2. Some figures can be improved

A. Figure 6 quality;

B. Figure 7 quality;

C. Figure 8 axis legends and units are very different from axis values.

D. Figures 10, 11 and 12 are missing some axis legends and units. Also, the caption is not in English in Figure 10 (b)

E. Some Figures still have greyed-out parts from Matlab.

 

3. Equation 10, 11, 12, 13 and 14 are strange, I would suggest authors to verify captions and indices.

4. Regarding the keywords I would like to suggest the authors to very them:

"ship intention identification; AIS; RANSAC; Bayesian framework; YOLO"

 

5. The sections after the conclusion (“Supplementary Materials”, “Author Contributions”, and so on) have not been filled out.

 

6. In the introduction, the cited amounts, from IMO for instance, should be included in the references.

 7. In related works the paragraph “Some other works are also of great research value and contribute to the prediction of ship intentions.” is too short.

 

When evaluating the work results, I would like to suggest few improvements:

 

1. The authors describe the dataset used for classification training, however, the YOLO performance is not evaluated. A confusion matrix would be interesting to better understand the method performance.

 

2. The results from Figures 10 and 11 are just shown without any further explanation. The authors need to elaborate further.

 

3. Author claims in the abstract that “A large number of experiments on two real data sets show that our proposed method is superior to the most advanced method for ship intention identification at intersections.” However, the results do not present this clearly.

 

4. I miss a comparison with the results of other similar works. 

 

5. What would be the future development from this research?

 

Author Response

We appreciate the thoughtful comments and questions regarding our manuscript entitled “Ship intention prediction at intersections based on vision and Bayesian framework” by Qianqian Chen, Changshi Xiao, Yuanqiao Wen et al., submitted for publication in Journal of Marine Science and Engineering. We have significantly benefited from your comments and improved our manuscript accordingly.  Our revised manuscript and our answers to the reviewers’ comment & questions are hereby submitted for your consideration. To be easily recognized, the answer to each comment and the revision in the manuscript were marked in blue color. Please contact us if you have any further question.

Author Response File: Author Response.pdf

Reviewer 4 Report

Dear Authors,

The papers' scope is to introduce an algorithm based on the image sequence and radar information with the aim to identify the navigation intention of ships at intersections.

The paper is interesting and promising, however, the scientific soundness and maritime English language are low. I suggest English editing. Additionally, in the paper are still fragments of the template.

  • When you discuss the AIS position accuracy, I suggest mentioning the AIS sampling rate, which varies depending on ships' speed, ROT, etc. 
  • Please discuss the possible weak points, such as data delay, meteorological impact, dense traffic.. 
  • The real-ship experimental data acquisition is not clear. Please clarify the experimental design.
  • The control (validation) of predicted results is not well discussed
  • In conclusion, I suggest mentioning future work. 

Best Regards

Author Response

We appreciate the thoughtful comments and questions regarding our manuscript entitled “Ship intention prediction at intersections based on vision and Bayesian framework” by Qianqian Chen, Changshi Xiao, Yuanqiao Wen et al., submitted for publication in Journal of Marine Science and Engineering. We have significantly benefited from your comments and improved our manuscript accordingly.  Our revised manuscript and our answers to the reviewers’ comment & questions are hereby submitted for your consideration. To be easily recognized, the answer to each comment and the revision in the manuscript were marked in blue color. Please contact us if you have any further question.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Thank you very much for including my comments in the article. I believe that the quality of the article has improved, but my doubts are still about the practical applicability of the model. It would be useful to add a paragraph in the article discussing this problem. For example, it would be good to know if the prediction distance is sufficient to undertake a collision avoidance manoeuvre? How is defined the point for which the coordinates of the vessel are determined? Are the dimensions of the vessel taken into account e.g. when determining the values given in figure 11? For the relatively short distances, the dimensions of the ship would be significant.

 

The description of the simulation results (4.3) is very unclear.

  • Figures 11 and 12 in part (b) show a velocity [km/h] but they are named trajectories.
  • The description of the figures refers to angular velocities, which should be given in e.g. [deg/min].
  • In figure 12 there is no part (c) that is mentioned in the caption
  • In the caption to figure 13, vessel 1 is mentioned twice

 

Additional comments:

  • It is not clear which figure the sentence “the area within the red dotted line” (line 314) refers to. It should start with a capital letter.
  • How the negative probability values shown in figure 8 (a) should be understood?
  • It seems that some of the trajectories marked in Figure 10 do not match the description (e.g. (b))
  • Scenarios e and f have the same description (lines 418-419)

 

Additionally, there are still editing errors in the article:

  • Wrong figure number (line 255)
  • Errors in the numbering of subsections (3.2 and 3.3 used twice)
  • Errors in the numbering of figures (lines 394)
  • Wrong numbering of figures 11-12 (lines 477-493)
  • Repeated sentence (lines 526-527)

Author Response

  1. For example, it would be good to know if the prediction distance is sufficient to undertake a collision avoidance manoeuvre? How is defined the point for which the coordinates of the vessel are determined? Are the dimensions of the vessel taken into account e.g. when determining the values given in figure 11? For the relatively short distances, the dimensions of the ship would be significant.

Answer: We appreciate the valuable comment by the reviewer. According to the regulations of the people's Republic of China on river collision prevention, the safety distance of ships with a length of more than 30m is 2km, and that of ships with a length of less than 30m is not less than 1km. Therefore, our algorithm can effectively avoid collision by predicting the ship intention within 2km. And in this paper, the position coordinates of ships are measured by radar. At present, the ship size is not considered when predicting the ship intention, and the ship size and speed will be considered in the future work.

  1. The description of the simulation results (4.3) is very unclear. Figures 11 and 12 in part (b) show a velocity [km/h] but they are named trajectories. The description of the figures refers to angular velocities, which should be given in e.g. [deg/min]. In figure 12 there is no part (c) that is mentioned in the caption. In the caption to figure 13, vessel 1 is mentioned twice.

Answer: We appreciate the valuable comment by the reviewer. Figures 11, 12 and 13 have been proofread.

3.It is not clear which figure the sentence “the area within the red dotted line” (line 314) refers to. It should start with a capital letter.

Answer: We appreciate the valuable comment by the reviewer. The manuscript has been proofread and improved carefully. These revisions in the manuscript are highlighted with blue color. (line 314)

  1. How the negative probability values shown in figure 8 (a) should be understood?

Answer: We appreciate the valuable comment by the reviewer. Negative probability indicates the probability that the ship turns in the opposite direction.

  1. It seems that some of the trajectories marked in Figure 10 do not match the description (e.g. (b)). Scenarios e and f have the same description (lines 418-419)

Answer: We appreciate the valuable comment by the reviewer. Figure 10 has been proofread

6.Additionally, there are still editing errors in the article: Wrong figure number (line 255); Errors in the numbering of subsections (3.2 and 3.3 used twice); Errors in the numbering of figures (lines 394); Wrong numbering of figures 11-12 (lines 477-493); Repeated sentence (lines 526-527)

Answer: We appreciate the valuable comment by the reviewer. The manuscript has been proofread and improved carefully. These revisions in the manuscript are highlighted with blue color.

 

Reviewer 3 Report

Thank you for the review opportunity.  The revised version did not include the sections after the conclusion (“Supplementary Materials”, “Author Contributions”, and so on). So the editor may verify them before publication.

The authors have addressed my concerns and the paper can be accepted.

Author Response

Answer: We appreciate the valuable comment by the reviewer.“Author Contributions” and "Declaration of competing interest" have been added in the manuscript.

Back to TopTop