Multi-UAV Cooperative Target Assignment Method Based on Reinforcement Learning

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this paper, to overcome the problems of traditional distributed target allocation algorithms in terms of lack of target strategic priority, poor scalability and robustness, this paper proposes a proximal strategy optimization algorithm that combines threat assessment and attention mechanism(TAPPO). In general, the paper describes clearly and well organized. The obtained results are interesting. English is generally very good and need to be improved. But there are faintly problems need to explain.

1) In equation 1 and 2, the letters of the formula need to be explained, such as pi_{theta}, s_{t}, a_{t}, and AT should be corrected as A_{t}.

2) Self-attention is also used in many fields, it is normal, whether it can be improved?

3) The motivation of this research is insufficient. I think the authors should explain the background and motivation of this research in a detail. The main problems can be summarized in Introduction.

4) The introduce of dataset should be explained in this paper.

5) More performance indicators for TAPPO algorithm should be presented.

6) In the Reference, the format is different, it should be improved. For example, [2] and [3] are different.

7) The section of Discussion should be presented in this paper.

Comments on the Quality of English Language

1) In equation 1 and 2, the letters of the formula need to be explained, such as pi_{theta}, s_{t}, a_{t}, and AT should be corrected as A_{t}.

2）In the References, the format is different, it should be improved. For example, [2] and [3] are different.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The submitted manuscript presents interesting results of target allocation by integrating threat assessment and dynamic attention strategies. The authors propose a dynamic reward function to improve exploration and scalability, achieving an 85% win rate in simulations. This article is relevant within the scope of the “drones” journal. This reviewer has some points to suggest:

1. Authors should state the meaning of acronyms when they first appear in the text;

2. Acronyms, when introduced, should not be italicized;

3. Fig. 1 - there are some Chinese characters;

4. This reviewer suggests that authors put figures in vector format, thus ensuring quality. Some figures are pixelated, with low resolution;

5. Section 2.2. More details on how the attention mechanism can be used in the work must be provided. As it is within the section on related work, this reviewer suggests the authors insert a study of how it is being used in related work and within the scope of UAVs;

6. In Table 3, it is necessary not to present the time, as it is relative to computational capacity, together with the use or not of CPU/GPU. It is recommended to use instruction metrics such as FLOPS or computational complexity;

7. Check the capitalization of equation items, such as equation (7);

8. There is a lack of clarity regarding how the experimental procedure can be integrated into the framework. The authors should provide a pipeline diagram of the results, starting from the variable that is being observed, the reinforcement learning system, and which variables will be used for the environment.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

In this paper, authors have provided a good revised version. In general, the presentation of this manuscript is good. However, there exist some minor presentation issues.

1. In page4, x represents the input sequence data, where is x presented?

2. The comparision of mainstream algorithms should be improved, just one index?

Comments on the Quality of English Language

In page4, x represents the input sequence data, where is x presented?

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have addressed the reviewer's concerns. After the new information, this reviewer has additional points:

1. After the citation in the text, the authors need to put a space;

2. In-line citations need to be fixed when there are more than three authors: For example: "Peter et al. [3]";

3. Concerning the FLOPS discussion, how can the proposed method be used in physical systems if it is more expensive than the methods found in the literature? Is it necessary to use better-embedded systems? 

4. The abbreviation and meaning need to follow the same pattern. Use the meaning and in parenthesis the abbreviation;

Comments on the Quality of English Language

There are some typos and ponctuation errors.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Many thanks for addressing the reviewer's concerns. This reviewer has no further comments and congratulates the authors for their research.

Author Response

Thank you very much for your valuable comments on my manuscript. I wish you good work and good health.