Thermal and Visual Tracking of Photovoltaic Plants for Autonomous UAV Inspection
, Carmine Tommaso Recchiuto 1, Jacopo Calla 2, Paolo Scuteri 2 and Antonio Sgorbissa 1,*
Round 1
Reviewer 1 Report
The article proposes a novel approach using an UAV with both an RGB and a thermal camera for PV module tracking. The UAV moves along PV module rows at a lower height and inspects them in a boustrophedon way by ignoring “empty" areas. Several suggestions are listed below.
1. In table 1, authors mentioned five different daylight conditions. Please describe how different they are. In addition, please describe the range of daylight conditions that the proposed system can work properly.
2. It’s suggested to construct a table which compares “average error, standard deviation, and navigation error” versus “thermal camera, RGB camera, and both camera”. Furthermore, please address the advantages of the proposed approach by using this table.
3. In Line 531, a constant velocity, 0.6 m/s, is mentioned. In Line 634, the speed is increased to 1.2 m/s. Please explain the change and address how the speed affects the experimental results. Authors mentioned the inspection of PV plants is a time-consuming task, which challenges the endurance of an UAV. Is a tether-powered UAV helpful to execute this mission? Please refer to https://doi.org/10.3390/app112411887.
https://ieeexplore.ieee.org/abstract/document/7443016/
https://ieeexplore.ieee.org/abstract/document/7078764
4. In Figure 9, 12, and 15, authors show the parameters for the thermal camera, RGB camera, and both cameras. Please not only show them but also compare them; thus, readers can understand the advantages of the proposed approach.
Author Response
1. In table 1, authors mentioned five different daylight conditions. Please describe how different they are. In addition, please describe the range of daylight conditions that the proposed system can work properly.
ANSWER: The required information has been added in Section 7.1 and Table 1
2. It’s suggested to construct a table which compares “average error, standard deviation, and navigation error” versus “thermal camera, RGB camera, and both camera”. Furthermore, please address the advantages of the proposed approach by using this table.
ANSWER: Tables 2 and 3 have been added, which summarize results in simulation and in the real world. Tables have been used to compare the RMSE.
3a. In Line 531, a constant velocity, 0.6 m/s, is mentioned. In Line 634, the speed is increased to 1.2 m/s. Please explain the change and address how the speed affects the experimental results.
ANSWER: A comment has been added in Section 7.3
3b. Authors mentioned the inspection of PV plants is a time-consuming task, which challenges the endurance of an UAV. Is a tether-powered UAV helpful to execute this mission? Please refer to https://doi.org/10.3390/app112411887.
https://ieeexplore.ieee.org/abstract/document/7443016/
https://ieeexplore.ieee.org/abstract/document/7078764
ANSWER. This is a very interesting suggestion that may be worth exploring. However, we are afraid that, due to the dimensions of the PV plant to be covered in autonomy, it looks complex to use this solution. Also, we think this is beyond the scope of the article.
4. In Figure 9, 12, and 15, authors show the parameters for the thermal camera, RGB camera, and both cameras. Please not only show them but also compare them; thus, readers can understand the advantages of the proposed approach.
ANSWER: The errors in the parameters' estimate have been computed, inserted in Table 2, and used for a quick comparison.
Reviewer 2 Report
This article proposes a new solution to planning Unmanned Aerial Vehicles (UAVs) to check the solar panel in photovoltaic (PV) plants. Its innovation comes from a conventional image processing method for PV module tracking based on an autonomous UAV with an RGB and a thermal camera. Finally, the paper verifies the effectiveness of the new method through simulations and physical verification.
The authors' study is interesting, and the proposed methods are applicable. It is a topic of interest to the researchers in the related areas but the paper could do some improvements before acceptance for publication. Comments are listed below:
(1) The contribution of the paper is not clearly displayed. The abstract only shows how the UAVs work in PV module tracking, but the importance of non-GPS sensors in this research is not highlighted, for example, image processing. Then, the authors are suggested to depict more characteristics of their work in the introduction section and summarize the important results obtained in the experiment.
(2) Some figures in the article do seem not clear enough to read. It is recommended that the authors modify the figure format, text size, typesetting, etc., and use high-resolution figures in the article, e.g Figure 1. Then, it is suggested that the author add more pictures of the gazebo simulation model or physical experimental equipment to enrich the content of the article.
(3) There may be a lack of theories on the path-planning controller and image splicing but a long space is used to describe the process of the experiment. In addition, in the study of PV module tracking based on an autonomous UAV， we hope to know more information about the accuracy of defect detection and time of the PV module tracking task.
(4) Please correct the errors in English grammar, spelling, and sentence structure so that the goals and results of the study are clear to the readers..
Author Response
1) The contribution of the paper is not clearly displayed. The abstract only shows how the UAVs work in PV module tracking, but the importance of non-GPS sensors in this research is not highlighted, for example, image processing. Then, the authors are suggested to depict more characteristics of their work in the introduction section and summarize the important results obtained in the experiment.
ANSWER: The abstract and the introduction has been revised according to the reviewer’s suggestions.
(2) Some figures in the article do seem not clear enough to read. It is recommended that the authors modify the figure format, text size, typesetting, etc., and use high-resolution figures in the article, e.g Figure 1. Then, it is suggested that the author add more pictures of the gazebo simulation model or physical experimental equipment to enrich the content of the article.
ANSWER: Figures have been revised and/or resized.
(3) There may be a lack of theories on the path-planning controller and image splicing but a long space is used to describe the process of the experiment. In addition, in the study of PV module tracking based on an autonomous UAV， we hope to know more information about the accuracy of defect detection and time of the PV module tracking task.
ANSWER: Defect detection is not the focus of this article. Comment has been added in the Introduction and the State-of-the-art to clarify this.
(4) Please correct the errors in English grammar, spelling, and sentence structure so that the goals and results of the study are clear to the readers.
ANSWER: We corrected errors in English grammar and revised the sentence structures in several places.
Reviewer 3 Report
1. In the Abstract section, please specify the model or method used in this paper, and the effect or improvement level of the corresponding experiment.
2. In the Introduction section, the explanation of academic value is not deep enough. Please list the main contributions of this paper in detail.
3. Based on the current research progress, please refer to the following or other literatures to further elaborate the research of swarm intelligence algorithm and multi UAV collaboration mechanism on this issue: ‘Dynamic Reallocation Model of Multiple Unmanned Aerial Vehicle Tasks in Emergent Adjustment Scenarios’ and ‘A review on representative swarm intelligence algorithms for solving optimization problems: Applications and trends’.
4. The prior knowledge mentioned in Formula (1) and the construction and solution of Formula (5) need further explanation.
5. What is the optimization algorithm used in the threshold optimization time in Section 7.1?
6. For the future research work in the Conclusion section, the description is too redundant, please summarize the contents.
Author Response
1. In the Abstract section, please specify the model or method used in this paper, and the effect or improvement level of the corresponding experiment.
ANSWER: The abstract has been updated accordingly to the reviewer’s suggestion.
2. In the Introduction section, the explanation of academic value is not deep enough. Please list the main contributions of this paper in detail.
ANSWER: The introduction has been revised by adding more details.
3. Based on the current research progress, please refer to the following or other literatures to further elaborate the research of swarm intelligence algorithm and multi UAV collaboration mechanism on this issue: ‘Dynamic Reallocation Model of Multiple Unmanned Aerial Vehicle Tasks in Emergent Adjustment Scenarios’ and ‘A review on representative swarm intelligence algorithms for solving optimization problems: Applications and trends’.
ANSWER: A citation has been added in the Conclusions when mentioning teams of drones.
4. The prior knowledge mentioned in Formula (1) and the construction and solution of Formula (5) need further explanation.
ANSWER: Two comments have been added concerning this: one comment is located near formula (1), and another comment is near formula (5)
5. What is the optimization algorithm used in the threshold optimization time in Section 7.1?
ANSWER: In Section 7.1, we added a reference to section 4.3, where the optimization procedure and algorithm are described.
6. For the future research work in the Conclusion section, the description is too redundant, please summarize the contents.
ANSWER: The “future work” part has been revised by moving some important information earlier in the section.
