Robotic Multi-Boll Cotton Harvester System Integration and Performance Evaluation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Review on the Manuscript:

Paper ID: agriengineering-2825494

Title:  Robotic Multi-Boll Cotton Harvester System Integration and 2 Performance Evaluation

 

I carefully read this manuscript. In this manuscript, a robotic multi-boll cotton harvester was developed, and its harvesting performance was tested.  I recommend the acceptance of this manuscript for publication, after the authors make some recommendations:

1.       The scientific contribution of the authors should be more clearly stated in introduction;

2.     The authors must emphasize, in the Introduction, the novelty of the article and implicitly the improvements brought by this article compared to the achievement in the specialized literature.

3.     An observation, valid for all the equations in the article: when after the equation, you continue with a sentence starting with a lowercase letter, you must put a comma, and when after the equation you continue with a sentence with a capital letter, you must put point.

4.     For the use of the PID controller (proportional, integral and derivative), with the command function from equation (6), it is necessary to present more information. 5.      The results presented in tables, pictures and graphs, being numerous, the authors should find a clearer and more concise form of presentation.  

I am convinced that it is useful for the manuscript to be included in the References section the following paper, which use similar procedures

 

i)               Advanced Control Subsystem for Mobile Robotic Systems in Precision Agriculture. International Journal of Robotics and Automation Technology, 9, 8–16, (2022). https://doi.org/10.31875/2409-9694.2022.09.02, https://www.zealpress.com/jms/index.php/ijrat/article/view/409/331

ii)              Distance Assessment by Object Detection—For Visually Impaired Assistive Mechatronic System. Appl. Sci. 2022, 12, 6342. https://doi.org/10.3390/app12136342, https://www.mdpi.com/2076-3417/12/13/6342#metrics

 

If the authors take into account all these corrections, then this manuscript deserves to be published.

10.01.2024

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

 

1.      Line 75 “In a study conducted in Reference [22],…” It is not proper to write the word Reference as it appears in this sentence. The word reference appears also in Line 82.

2.      Figures should be mentioned in the text first before being presented. Example Line 292 Figure 8 (b) is mentioned after being presented.

3.      Figure 12: Remove the words “Picking Ratio” on top of the figure.

4.   There are not enough statistics comparing the lab and field experiments. Standard deviation only is not enough. Here, you can do some analysis to conclude whether there is a statistical difference between the robot in 2022 and the new one.

 

5.  Lines 436 -440 are presenting results, it should not be in the discussion part. It is not common to present photos in the discussion part (Figure 13)

 

6.      The information presented in Table 3 caption does not match what is written in the text. Line 414 is written “Table 3 shows the average harvesting performances” while Line 425 “Table 3. Summary of the lab tests and the field tests.” Is contradicting.

Comments on the Quality of English Language

Some minor corrections required

Author Response

Please see the attachment. 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study uses YOLOv4 for boll detection, and with a processing time of 2.5 seconds per boll, the resulting accuracy rate is 57.1. This is a very low result. The YOLOv4 model was proposed in 2020. Currently, there are many improved versions of YOLO for more accurate results and lower calculation time such as YOLOv5, YOLOv6, YOLOv7, YOLOv8, YOLO-Nas. So why don't you test on these versions?

I understand that this is also the core of the cotton detection software that serves the system's cotton collection.

You should refer to this article: YOLO Series for Human Hand Action Detection and Classification from Egocentric Videos, https://www.mdpi.com/1424-8220/6/23/3255

Comments on the Quality of English Language

English needs to be improved in terms of spelling and grammar.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The work is clearly presented and of interest to the cotton engineering community. 

Minor editing for a few grammatical issues is suggested, such as:

Line 61 "All" should be "all"

Line 275 "on THE embedded"

1) The manuscript addresses the idea of robotic small-scale autonomous cotton harvest which would be a revolutionary change to the mechanical harvesting of cotton. 2) The research presented is an update on improvements to previously published autonomous cotton harvest systems with improved boll detection and harvest efficiency. 3) Comparison of the performance of 3 versions of the YOLO object detection software and performance on the microprocessor, improvements to the end effector and pull back reel are all specific improvements over previously published work. 4) The work as presented is complete and does not need improvement. 5) The conclusions are supported by the field trials of the new system and data presented. 6) References are complete and appropriate. 7) Figures are clear and add value to the manuscript.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Review on the Manuscript:

Paper ID: agriengineering-2825494-peer-review-v2

Title:  Robotic Multi-Boll Cotton Harvester System Integration and Performance Evaluation

 

 

In my opinion, taking into account the author's answers and corrections, I recommend the acceptance of the manuscript for publication. However, the final decision is up to the editor-in-chief. 

13.02.2024

Author Response

Thank you so much for providing your insightful comments and for the valuable efforts you have made to improve the manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you, I saw your response to my comments about using newer versions of the YOLO network. You answered me that when you train, YOLOv4 is the most effective. However, YOLOv4 was announced in 2020, so it has been 4 years now. Versions YOLOv5 and YOLOv6 were released soon after. So your research does not guarantee novelty and topicality. I have had quite a long process of working on versions of YOLO. Using YOLOv5, YOLOv6, YOLOv7, YOLOv8 is very simple and the training time is also short. Therefore I would like to see comparative results of Bong detection across versions of YOLO. This will make your research more convincing.

Comments on the Quality of English Language

Minor correction to English

Author Response

Comments from Reviewer 3

• Comment 1: Thank you, I saw your response to my comments about using newer versions of the YOLO network. You answered me that when you train, YOLOv4 is the most effective. However, YOLOv4 was announced in 2020, so it has been 4 years now. Versions YOLOv5 and YOLOv6 were released soon after. So your research does not guarantee novelty and topicality. I have had quite a long process of working on versions of YOLO. Using YOLOv5, YOLOv6, YOLOv7, YOLOv8 is very simple and the training time is also short. Therefore I would like to see comparative results of Bong detection across versions of YOLO. This will make your research more convincing.

 

Author’s response: Thank you for your comment. We agree with the fact that YOLOv4 is not the latest version of the YOLO network; however, this was a long research project that started in 2020, with different components of the study conducted at various times. We conducted experiments on cotton boll detection, robot development and control, autonomous navigation, and finally, we explored different cotton harvesting end-effector designs and implementations. The cotton boll detection component was the first to be implemented back when YOLOv4 was still the newer version of the YOLO network, and it was compared to YOLOv3. When testing the solution, we observed that the YOLOv4's 92% precision was sufficient for detecting the cotton bolls and for the system to determine the placement of the end-effector, as we only needed to know the majority of the bolls' locations (to cluster them and find the centroid). We found that the primary factor affecting the effectiveness of the solution was not boll detection, but rather the efficiency of the end-effector in extracting cotton bolls and the loss of cotton bolls dropping on the ground during harvesting. Therefore, changing the detection model was not a priority. We cannot test the newer versions of the model in the field within the current scope of the study.

 

The novelty of the study lies in the fact that no other research has studied harvesting multiple cotton bolls simultaneously autonomously. This is a novel idea that harvests multiple cotton bolls by moving a specially made end-effector based on the centroid of clusters of the detected bolls. Our solution was implemented and tested in both controlled and real cotton field environments. In future studies, we will consider newer detection models.

 

• Comment 2: Comments on the Quality of English Language, Minor correction to English.

Author’s response: Spelling and grammar were corrected.

Round 3

Reviewer 3 Report

Comments and Suggestions for Authors

I feel unsatisfied with the authors' answers. Although cotton detection is only an initial result of the system. However, this result is very important in the two-Bong collection system. Therefore, it needs the latest implementation and testing of YOLO.

Comments on the Quality of English Language

ok