Review Reports - A Multi-Drone System Proof of Concept for Forestry Applications

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents a comprehensive method to leverage multi-drone systems for efficient and autonomous forest mapping. This work combines positioning and mapping, collaborative multi-drone mapping, robust control and navigation, and swarm formation as a complete framework. Detailed comments are as follows.

This manuscript spends a lot of space to introduce the various modules, but the location of the main innovation highlights(DCL-LIO-SAM) is too late, which leads to the shift of the focus of the reading. Therefore, the article needs to adjust the way of organizing the writing.

Figure 11. Can you add comparative methods to express the superiority of your method, such as the time used?

Line263, This work seems to be a multi-tasking portfolio, and the mentioned work DCL-LIO-SAM is the innovation in your article. Can you list the enhancements compared to only DCL-SLAM and only DIO-SAM?

Line1063, What are the differences between these three types of UAVs in terms of total closed loop, inter-loop closure and intra-loop closure?

Line 340 wireless network modules, can you discuss the specific technical details of wireless networks, such as the frequencies used, communication protocols and data transfer rates, to make the parameters clear.

Line 794 What is the adgvantagement of the PSO compared with the others optimization algorithms, such as Ant Colony Optimization, Simulated Annealing and Genetic Algorithms? Do you have a comparative method to express it.

At last, I think you can emphasize the innovation of this paper, it may help your work be easier to understand.

Author Response

Comments 1:

Response 1:

In our paper, DCL-LIO-SAM refers to the DCL-SLAM framework using a front-end module based on LIO-SAM. While this integration was originally presented by the authors of DCL-SLAM and is not a novel contribution of this work, our primary innovation lies in combining state-of-the-art frameworks, including DCL-LIO-SAM, with the capabilities of the MRS system. Notably, we have made several improvements to DCL-LIO-SLAM, such as incorporating a GNSS factor into the Pose Graph Optimization (PGO) module, refining the loop closure process, integrating a global map service, and standardizing naming conventions and reference frames for compatibility with the MRS UAV System.

To address the identified concern, we propose to emphasize this combining state-of-the-art frameworks as the central contribution, ensuring it is clear to the reader from the start (Section 1.2 Objectives and 6. Conclusion).

---

Comments 2:

Figure 11. Can you add comparative methods to express the superiority of your method, such as the time used?

Response 2:

We would like to clarify that the intention of this figure is not to compare methods but rather to illustrate the progressive mapping of the environment by a single drone during our field experiment. Specifically, Figure 11 demonstrates the gradual construction of the map at four distinct moments as the drone explores the area. This visualization highlights how newly captured features are incrementally integrated into the overall representation, showcasing the real-time capabilities of the system rather than its performance relative to other methods.

As this study is focused on integrating various frameworks to create a cohesive multi-drone system, performance benchmarking against other methods was not within the scope of this phase of the project. Future work might aim to provide such comparative evaluations once the system has been fully validated and optimized for broader application scenarios.

To address the identified concern, we propose updating the manuscript text accompanying Figure 11 to explicitly clarify its purpose (L 1042 - L1049). We believe this revision will help readers better understand the figure’s context and its significance within the study.

---

Comments 3:

Response 3:

In Section 3.2.4, we outline the specific advancements and improvements made in integrating DCL-SLAM and LIO-SAM into the unified DCL-LIO-SAM framework. To reiterate, the key enhancements are as follows:

Integration of GNSS Factors: Unlike the original DCL-SLAM and LIO-SAM frameworks, our integration includes a GNSS factor within the Pose Graph Optimization (PGO) module. This addition enhances localization accuracy, especially in outdoor environments where GNSS data can complement LiDAR and IMU measurements to reduce drift; and
Global Map Service Integration: A significant addition to the framework is the inclusion of a global map service, which aggregates local maps from multiple drones into a unified representation. This integration facilitates efficient collaboration among drones in multi-robot operations and supports a shared understanding of the environment.

-----

Comments 4:

Line1063, What are the differences between these three types of UAVs in terms of total closed loop, inter-loop closure and intra-loop closure?

Response 4:

In our system, the UAVs are identical in both hardware and software configurations, including their sensors and SLAM frameworks. Therefore, the distinctions in total loop closure, intra-loop closure, and inter-loop closure are not based on differences in UAV capabilities but rather on their trajectories.

The ability to achieve inter-loop closure relies on the degree of overlap between the exploration areas of different UAVs. In contrast, the effectiveness of intra-loop closures depends on the characteristics of the environment being explored, which influence a UAV's ability to recognize previously visited locations.

We propose updating the manuscript with this explanation. (Line 1054 e 1059)

—

Comments 5:

Line 340 wireless network modules, can you discuss the specific technical details of wireless networks, such as the frequencies used, communication protocols and data transfer rates, to make the parameters clear.

Response 5:

The technical details are mentioned in the Section 4.1.1. “Technical specification”. The wireless communication in our system operates on the frequency bands, commonly used for Wi-Fi communication (802.11ac). Additionally, the drone includes an LTE Cat4 mobile router, enabling it to operate over greater distances by leveraging cellular networks. We use ZeroTier which is a software-defined networking (SDN) platform that enables the creation of virtual networks over the internet. The system uses Nimbro network protocol, designed for high-latency, low-quality connections and is optimized for reliable ROS message transport, supporting both topic-based communication and service calls.

—

Comments 6:

Response 6:

In this study, PSO was selected due to its simplicity for real-time parameter tuning in dynamic systems in order to reduce the time and effort required for manual tuning. Additionally, one of the authors has significant experience working with PSO, having conducted extensive research and published works on its application in various domains. The combination of PSO’s practical advantages and our expertise ensured it was the most suitable choice. Future work might aim to provide such comparative evaluations once the system has been fully validated and optimized.

----

Comments 7:

At last, I think you can emphasize the innovation of this paper, it may help your work be easier to understand.

Response 7:

As described in the first comment, we propose reorganizing the manuscript to prominently emphasize this integration as the central contribution, ensuring it is clear to the reader from the start.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1. Some mathematical symbols should be modified, e.g., the matrix should be written in bold font.

2. In Algorithm 1, how to obtain the fitness and update it with different iterations?

3. In table 5, it seems unclear of the formations A-F.

Author Response

Comments 1:

Some mathematical symbols should be modified, e.g., the matrix should be written in bold font.

Response 1:

The matrix representations, mainly in section 3.3.1, were corrected accordingly.

—

Comments 2:

In Algorithm 1, how to obtain the fitness and update it with different iterations?

Response 2:

To address the clarity of how the fitness is obtained and updated across iterations, we have updated three lines of the algorithm (Algorithm 1 -> Lines 8, 15 and 17). The fitness is calculated by evaluating the root mean squared error (RMSE) between the drone’s real output (i.e., its estimated position from the SLAM framework) and the desired target position, as presented in equation 12.

—

Comments 3:

In table 5, it seems unclear of the formations A-F.

Response 3:

The formations referred to in table 5, are relative to variations in swarm formations over six distinct moments (A,...,F) and overall trajectories of each individual drone. It was added in the table caption a reference to the Figure 14a), to be more clear.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

I would like to thank the authors for their work on this article.

The authors present a novel and robust methodology for the survey and mapping of forest environments based on the use of drone swarms.

This work is of great scientific interest due to the potential applicability in different forest environments. However, this work needs to improve several aspects detailed below, but which are mainly focused on the order of the text and other aspects that need to be improved and clarified, especially in the extensive use of acronyms, the use of terms that may not be in accordance with the research, the justification and description of the study site, the improvement of the scientific language, the improvement of the discussion section and consequently of the conclusions.

These adjustments will make the text more orderly, much easier to understand and much more robust from a scientific point of view.

Comments and recommendations:

L 3-4: If these are acronyms, they cannot be presented without a prior definition, please review along the text.

L 16: Utilization?

L 83-84: Here the authors could include some examples of the fields in which these advances have been obtained, especially because they are not going to be mentioned in detail later, so having an idea before continuing the reading helps a lot to the understanding, it is not enough just to add a reference of another author who has already made a review.

L 96: This is the first time that the acronym LiDAR is used in the text, so it must be defined before.

L 107: LiDAR must be defined before.

L 122-123: The authors could add how it is possible for drones to perform these tasks, or at least give an example of the sensors used to perform this function.

L 134: Although the objective of this section is to address the previous work of other authors in drone-based forestry research, it is very difficult to maintain a reading flow due to the large number of acronyms present, in addition, some acronyms were presented in the text without being previously defined. I recommend the authors to work on the organization of this section, perhaps adding a table with the acronyms and definitions would help to improve the comprehension of the text.

L 135: LiDAR must be defined before.

L 136: Localization?

L 316: The authors could provide examples of forestry operations where such applications will be useful.

Figure 1: This figure is somewhat confusing, there are some lines like the black segmented lines, which are not clear if their function is to connect, group or separate, please clarify this figure.

L 959: The authors should rewrite this section, as there are numerous errors and lack of information, it is one of the poorest sections of the article. It is essential that the choice of the study area is well justified and described in the most technical-scientific way possible. It is not possible to write “with slopes that present unique challenges”... in a scientific publication I would expect to read “with slopes varying from x% to y% and/or an average slope of z%”, as well as other suggestions detailed in the following comments.

L 961: The authors should better detail the topographic characteristics of the site.

L 964: This is not a scientific way of describing slope. Percentages or degrees should be specified. Furthermore, in a 100 meter section, a difference of 5 meters represents a linear slope of 5%, is that slope a unique challenge for data collection and drone operation?

L 967: How did the authors determine that there is a significant biomass, was it previously measured, if it was previously measured, what is the value?

L 969 - 971: This is not clear, since the application of these technologies in forestry operations is mentioned above, but throughout the text no details or examples of forestry operations are given. This term is commonly used to refer to the field of silviculture and the operations that are developed in that field (plantations, harvesting, timber transport, road design, etc), so it is not clear to me if the authors are referring to that type of forestry operations. If so, I would expect the selected site to be a place where such operations take place, that could further increase the interest in this work. If this is not the case, I suggest to work in a better way on the concepts used and the justification of the study area, to avoid confusion and clarify the objective of the application of this methodology.

L 1018: The authors could improve this section, although the results are explained, the discussion is a bit general and only a few references were chosen for comparison with other works (also in a general way). As I mentioned above, it is not clear the target field of this research, since it talks about forestry operations, but later it is not explained which ones, besides the study area does not seem to be a place where this type of operations are developed, and the discussions do not talk about how this technology can be useful in this field. I suggest to work more on these details to increase (even more) the interest of this research.

L 1264: What kind of forestry management?

Author Response

Comments 0:

These adjustments will make the text more orderly, much easier to understand and much more robust from a scientific point of view.

Response 0:

We appreciate your suggestions and would like to highlight that, in response to feedback from other reviewers, we have implemented several adjustments to improve the clarity and readability of the manuscript. We will revise the manuscript to ensure that all acronyms are clearly defined upon first mention. In addition, we have inserted a list of acronyms in the paper.

—

Comments 1:

L 3-4: If these are acronyms, they cannot be presented without a prior definition, please review along the text.

Response 1:

The acronym designation was added, more specifically LiDAR, SLAM, LIO-SAM and DCL-SLAM.

—

Comments 2:

L 16: Utilization?

Response 2:

Since we define British English (en-GB), utilisation is the correct form.

—

Comments 3:

Response 3:

It was added the specific domain of the actuation of this work, that is focusing particularly on forestry management, with an emphasis on forestry inventory (e.g., cataloging plants and fruits, assessing tree and vegetation density, in a certain location). L 86-89

—

Comments 4:

L 96: This is the first time that the acronym LiDAR is used in the text, so it must be defined before.

Response 4:

The LiDAR acronym was initially defined in the abstract section, L 4.

—

Comments 5:

L 107: LiDAR must be defined before.

Response 5:

The LiDAR acronym was initially defined in the abstract section, L 4.

—

Comments 6:

L 122-123: The authors could add how it is possible for drones to perform these tasks, or at least give an example of the sensors used to perform this function.

Response 6:

Added the technology used to assess the tasks referred in the text. L 127-128

—

Comments 7:

Response 7:

It was created a “Acronyms List” (L 1312) as suggested, and all section 2.2 was revised and improved to make it more clear to understand.

—

Comments 8:

L 135: LiDAR must be defined before.

Response 8:

The LiDAR acronym was initially defined in the abstract section, L 4.

—

Comments 9:

L 136: Localization?

Response 9:

Since we define British English (en-GB), localisation is the correct form.

—

Comments 10:

L 316: The authors could provide examples of forestry operations where such applications will be useful.

Response 10:

Added a description reinforcing the domain of application, L 321 - 324 .

—

Comments 11:

Figure 1: This figure is somewhat confusing, there are some lines like the black segmented lines, which are not clear if their function is to connect, group or separate, please clarify this figure.

Response 11:

Figure 1 was revised, by using the same color to the segmented lines it seems to be more clear that it belongs to the same group, as a sub-section. Also the main titles of the groups were modified to bold format.

—

Comments 12:

Response 12:

This section 4.3.1 was revised and improved in a way, it is more clear the choice of the study area. In summary, it was used the official laws of reforestation guidelines, where it is defined that spacing between trees of 5 meters minimum (https://files.diariodarepublica.pt/1s/2018/01/00901/0000200004.pdf Artigo 4º 1a), and the expected number of trees in the flight area of 0.354 hectares is approximately 144, in our test site we have around 31 trees, representing 22% of expected forestry density under these guidelines, making it adjusted for initial deployment tests.

—

Comments 13:

L 961: The authors should better detail the topographic characteristics of the site.

Response 13:

A more detailed text was written to make clear the topographic characteristics presented after mentioning it. L 964

—

Comments 14:

Response 14:

The sentence was revised, indeed the slope is not the challenge itself neither and relevant metric to be considered, but the morphology of the obstacles and density, in this case vegetation and trees, it was reinforced in the text.

—

Comments 15:

L 967: How did the authors determine that there is a significant biomass, was it previously measured, if it was previously measured, what is the value?

Response 15:

The authors aimed to highlight the biodiversity of the experimental area, characterized by various biomass species with distinct physical attributes, such as diverse tree branch topologies, varying leaf densities, branching spacings, and ground vegetation types. While the statement was not intended to imply a quantified measurement of biomass, it reflects the diverse structural and ecological features of the area that make it suitable for multi-drone mapping experiments. We acknowledge the importance of providing precise data and will clarify this in the manuscript to avoid any misinterpretation.

—

Comments 16:

Response 16:

The paragraph was revised emphasizing the objective of the application, L 977

—

Comments 17:

Response 17:

At this point of the revision, after all previous suggestions the authors believe that is clear the objective of this work and consequently the reason for these results.

—

Comments 18:

L 1264: What kind of forestry management?

Response 18:

Added further comments to describe the kind of forestry management. L 1290 - 1292

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you very much to the authors of this work for their efforts. They addressed my suggestions appropriately and improved the quality of the document.