Design of an Autonomous Cooperative Drone Swarm for Inspections of Safety Critical Infrastructure

Round 1

Reviewer 1 Report

The article is written in clear language and uses current sources.

 

I find it difficult to say that the article is well organised. Since it is a design article, too long and irrelevant information was given. Concepts were mentioned rather than a task, hardware design, or software, which made the article longer than it needed to be. I think the concept written in the project content is included here as it is needed. The paper contains a long introduction as if it were a book chapter. Then the methods section was concluded with a potpourri of short/summary explanations of many research problems. I think there was an attempt to pack too much into one paper and of course not everything was presented properly.

 

The explanation of the contributions in the definitions of the paper is actually sentences that are repeated. It should be highlighted as more specific assertions or contributions.

 

Although the literature review is up to date, there was not enough discussion and comparison. I consider this to be a major shortcoming. It is strongly recommended that one or more tables be included highlighting differences, contributions, and innovations compared to existing papers.

 

For a reader-friendly approach, a summary flowchart or descriptive drawing should be placed at the beginning of the second chapter and the materials and methods sections. The hardware section can show what the connections are and what they look like. I think that the placement of the camera and sensors is very important. It is a confirmation for each added finding. It is not possible to test the hypothesis of the experimental design by a researcher. For this reason, it is necessary to present/share and discuss more satisfactory image or video evidence.

 

Issues of communication, formation, etc. will be addressed only at the concept level. Is it really being executed? Has the system been activated? Is it just a matter of acceptance?

 

The manuscript in its present form has serious shortcomings and is not yet scientifically mature. It may be suggested that the article be resubmitted after maturation and demonstrable applications in place of the concept, but it is not appropriate to publish it in its present form.

Author Response

We would like to sincerely thank you for the careful reading of our manuscript and for the constructive comments. We recognize that the paper has not had the scope and focus as one may have expected. Purposely, our aim with the paper has been on the design aspect and to provide a fairly complete view of our drone system. Consequently, the paper may be perceived as long. In the following we highlight the changes made to accommodate your comments.

1. Length of the paper:  To shorten the paper we have decided to remove some sections. From the paper we have removed “3.1 Requirements”, “6.1 Network architecture”, “7.8 Drone Charging Protocol” and merging of the Communication sections. The two former topics are removed to focus on the design aspect and the topics will be deferred to future paper publications.
2. Shortening of the introduction: The introduction section has been rewritten to be shortened. The section has been split into the part that introduces the core topic and the contributions and the Related Work (new section). The description of the overarching use case has been moved to the introduction section. Overall the new introduction section is shorter than in the first submission of the manuscript.
3. The explanation of the contributions of the main contributions of the paper has been rephrased in Section 1.
4. We disagree with the reviewer that a summary flow chart is strictly needed in the beginning of the second chapter considering also the length of the paper. However, we have elaborated on the description of the paper outline in Section 1: “Introduction”.
5. To give more details on the hardware connections and what they look like, Figure 3 was introduced. The figure shows conceptually how different hardware submodules are put together in the design.
6. We have extended the discussion of the related work to provide more clarity on the comparison between our study and what we found in the literature.
7. We agree that video demonstrations for this type of research are good means of validation. The first submission contained a link to one video demonstration. This video demonstrated operational aspects of the UAS. We have added two more videos demonstrating the drone hardware and perception system and drone formation flying, respectively.
8. Concerning elements addressed only at the concept level we have decided to remove the parts concerning coordinated charging as this is currently only validated in simulations. 
9. We agree with the reviewer that the manuscript can be improved and have updated it accordingly. However, we disagree with the reviewer that the manuscript is not scientifically mature. From a UAS system point of view we provide completeness linking to references to specific research published from team members on specific scientific contributions linked to our UAS, for example references [15][24][26-29][38] and [50-51] deal with different part on the design including their validation.

Reviewer 2 Report

The authors proposed a design of an autonomous cooperative drone swarm for the inspection of critical infrastructure. This manuscript is well-organized and written clearly. However, I want to give several comments to improve this draft as follows:

In the abstract, the authors should describe the methodologies used for design and the key findings of experimental results.

It is amazing that drone services in healthcare and transportation are not mentioned at all.

It is better to explain why the authors choose the type of communication for each service, including drone-to-ground, drone-to-drone, and drone-to-cloud.

The authors should give some limitations and future works of this study.

Author Response

We thank the reviewer for the insightful comments. The paper manuscript has been updated accordingly. Below we summarize the updates related to the specific comments given by the review:

1. We have updated the abstract of the paper to highlight the methodology and the key output of the paper.
2. To also address applications in health-care and transportation, we have added reference [10][11] and [12]. 
3. We have introduced a new table (Table 1) to summarize and reason over the choice of communication technologies for the UAS.
4. A new Section 8.6 was added to briefly address limitations. This also underlines the mentioned future work to raise technology readiness level of the autonomous drone swarm and the UAS as a whole.

Reviewer 3 Report

1- The manuscript can be checked by native English speaker

2- I suggest to the authors to highlight (objectively) the innovation of this proposal in relation to literature.

3-The wider interpretation of the results can be added for the better understanding of the analysis.

4-      In methodology and Experiential Results section no explanation about drone there, please make it clear.

5-Paper structure need revision to make “reader” friendly, e.g. Authors jumped from experiment setup to performance evaluations without a clear focus on “main results”. a dedicated section for “main results is needed”.

6- The performance section, which is very challenging to see the technical contribution in a clear way.

7- It is not enough to judge the model performance simply from the viewpoint of drone, the data dynamics and many factors system efficiency may need to be considered in the modeling process.

8-   The references are old and recent references should be used

9- There are a number of references present in the list of references and they were not citied in the body of the study

1-     Enabling drone services: drone crowdsourcing and drone scripting, M Alwateer, SW Loke, N FernandoIEEE access 7, 110035-110049

2-      Drone services: An investigation via prototyping and simulation, M Alwateer, SW Loke, W Rahayu2018 IEEE 4th World Forum on Internet of Things (WF-IoT), 367-370.

Author Response

We would like to thank the reviewer for the careful reading of our manuscript and for the comments. 

1. The manuscript has been thoroughly checked for grammar and spelling mistakes.
2. The explanation of the contributions and the key innovation of the paper has been rephrased and highlighted in Section 1.
3. We have included a new section “8.6 Limitation of This Research” to bring a wider perspective to the results of our study. In addition, the Conclusion section has been updated to emphasize more on main results and wider impacts.  
4. To give more details on the hardware connections and what they look like, Figure 3 was introduced. The figure shows conceptually how different hardware submodules are put together in the design.
5. The paper is intended to be reporting on design aspects and we have therefore chosen to introduce the design of individual subsystems of the UAS i.e., the drone hardware, the communication subsystem, the most important algorithms. Hereafter follows validation of selected parts of the design. We believe that the conclusion of the paper summarizes the main results of the study.
6. The paper does not have a specific “Performance” section. The main technical contributions are summarized in the “Introduction” section and elaborated throughout the paper.
7. We agree with the reviewer that data dynamics is an important aspect. Our approach has been to rely on the Robot Operating System (ROS) with the MAVlink protocol for the key data distribution mechanism. This is a widely accepted approach in the robotics community and we have therefore considered analysis of data dynamics beyond the scope of this paper.
8. We disagree with the reviewer’s comment that references are old. Looking at the list more than 40% of the paper references are from 2021 or 2022 in the first submission.
9. We recognize that the LaTeX bibliography file (.bib) contained one reference that was not on purpose cited in the manuscript. We have cleaned up the .bib file.
   We find that references proposed by the reviewer have only marginal relevance to our study and have therefore not been cited. The first paper is a concept study in contrast to our work that reports on a design for a specific drone application. The second paper describes an interesting prototype for a fly-in-fly-out demand edge drone service. Although the paper embeds coordination between drones the paper does not sufficiently relate the application to inspections which is our focus.

Reviewer 4 Report

In this paper, the authors have reported on the design of a UAS to support safety-critical infrastructure inspections. The system design is based on an analysis of requirements and the concept of operation for the inspection. This system represents a technology frontier for the use of drones. The presentation of the contents are systematic and clear and it is well written. No obvious problems are detected and so it is suggested for acceptance.

Author Response

We thank the reviewer for the supportive comments concerning our submission.

Round 2

Reviewer 1 Report

The authors made notable improvements to the study. I'm also glad that the evidence is presented as a video. However, I think the study still needs improvement in terms of results and discussion. The presentation of the images processed by the aircraft at the scene of the incident is insufficient. It would be good to present the images processed inside the device (whether simulated or real) in the findings. It will be practical and interesting for the reader to present the study in a summary table that differs from the existing studies. The authors should clearly explain to the reader why a smartphone, autonomous ground vehicle, blimp or IoT device is not used instead of the proposed method. The pros and cons here should be reflected in the text realistically. Why weren't the advantages of systems such as blimp and balloons, such as energy saving or the fixed wing type vehicle scanning a wider area faster?

Author Response

We would like to sincerely thank the reviewer for the constructive comments. Those comments are all valuable and very helpful for revising and improving our paper. We appreciate the acknowledgment of the improvements made to the manuscript after the first round of reviewing.  

In the following, we highlight the changes made to accommodate the new comments made.

1. Examples of aircraft image processing can be found in the paper. The cable identification and tracking system is using a computer vision algorithm to estimate the cable orientation as described in Sec. 7.1. Fig. 6 shows the data created by the cable identification and tracking system. In addition, Fig. 10 and also video-1 show examples of images created by the drone vision system. For the fault identification with AI, Fig. 18 represents two examples of images used by the algorithm. Video 3 shows inspection images - simulated as well as real ones - taken by the drone. In addition, we have added two image examples demonstrating the identification of a fault insulator on a powerline cable in Figure 19. 
   
   
2. Summary (table) of the difference wrt. other similar studies: In this paper, we focus on a UAS design from a holistic viewpoint. We have not found papers that report on design aspects with similar breadth. As we align our design with several known concepts, one could argue that for comparison on a component level. However, we believe that this is becoming too exhaustive and we consider this to be beyond the scope of this paper. Nevertheless, the suggestion is good and relevant but should be deferred to a literature survey paper potentially becoming a guide for future UAS designs. 
   
   
3. “why a smartphone, …or IoT device is not used instead”:  We believe that IoT devices and smartphones can be applied in civil infrastructure inspections where monitoring of vibrations is the focus. In contrast, our application focuses on non-invasive photogrammetric inspections. We, therefore, do not see the relevance of such a comparison. 
   
   
4. Comparison with other types of vehicles: We have extended the related work section, including the addition of Table 1, to compare the different types of robotics inspection platforms and to reason over the choice of the multirotor design. The comparison includes ground vehicles, fixed-wing and rotor-wing vehicles as well as the robotic blimp. The analysis has led to the addition of 6 new references. 

Thank you very much. 

Reviewer 3 Report

Paper should recheck again by native English speaker

Author Response

We thank the reviewer for the support. The paper has now been carefully reviewed again by several professionals proficient in English.