Enhanced Visual Detection and Path Planning for Robotic Arms Using Yolov10n-SSE and Hybrid Algorithms

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The "Materials and Methods" section seems too long and overloaded with descriptions of small details (especially the description of filtering parameters, Alpha-shape, etc.).

2. The authors need to provide a comparative analysis with other modern dimensionality reduction methods or alternative hybrid planning schemes.

3.  The description of novelty is vague in places - the authors could have more clearly formulated what exactly is fundamentally new compared to existing robotic solutions.   4. The statistical significance of the gains is not always clearly stated in the comparisons. This reduces the objectivity of conclusions about the advantages of Yolov10n-SSE.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The reviewed article corresponds to the thematic areas of the Agronomy journal. It is devoted to the current direction of development of information technologies for agricultural purposes, namely, improving visual detection and path planning for robotic arms using Yolov10n-SSE and hybrid algorithms. The authors' research is characterised by scientific and applied value, but there are certain comments to the article that need to be taken into account and corrected by the authors.

1. Throughout the text of the article, ‘?’ is displayed in square brackets instead of reference numbers, which makes it impossible to analyse the quality of citation of well-known works. They are also not numbered in the list of references.
2. Links to figure numbers in the text of the article are also displayed with ‘?’.
3. The introduction section in the author's version is very generalised. It is recommended that the authors modify the structure of the material presentation. The introduction should state the relevance of the research problem, analyse the known results according to certain criteria (practical purpose, technologies used, effect achieved, etc.), summarise the research gaps in the known results, and justify the purpose and objectives of this article.
4. Why did the authors include Figure 1 in the introduction? In my opinion, this figure belongs in the Materials and Methods section.
5. Line 88: The authors state that the graphic images were obtained under different lighting conditions. However, they do not specify which ones.
6. It is also advisable to provide the growth stages of the pineapples for which these images were obtained.
7. The principle of operation of the structural diagrams in Figs. 3-5 and their purpose in the author's study should be described in more detail.
8. Figure 6: Quality should be improved (sharpness, colour and contrast).
9. Subsection 2.4 should describe the technical and functional characteristics of the robotic arm in more detail. It is also advisable to provide graphic representations of it.
10. It is recommended to supplement Algorithm 1 (page 7) with a flowchart to improve its readability and reproducibility.
11. A reference to the literature source of formulae (1) should be added.
12. Table 2 shows a comparison of the performance of different models. However, it would be desirable to supplement this data with estimates of the standard deviation for the indicators (Precision, Recall, AP, F1). Without the standard deviation, it is impossible to assess the statistical significance of the models, especially with the small deviations observed in the authors' calculations.
13. In subsection 3.2.3, the authors present the results of testing and comparing the algorithms (Artificial Potential Field Algorithm and Improved RRT* Algorithm) for planning 3D trajectories. However, it should be noted by what criterion the test maps were selected and how well they correlate with real conditions.
14. The authors provide Fig. 11 as evidence of the effectiveness of using SE, SPConv and EMA mechanisms. However, the graphical interpretation in Fig. 11 is not accompanied by quantitative estimates (e.g., the average value of the Intersection over Union metric), which would allow for an unequivocal statement that the spatial accuracy of object detection has improved.
15. It is necessary to add a Discussion section, which should indicate the scientific novelty and practical value of the research results in comparison with previously known ones, as well as the prospects for the development of the results obtained.
16. The authors' research focuses on the type of crops, pineapples. Therefore, it is appropriate to reflect this in the title of the article.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Overall, the article is well-organized and structured. It employs a suitable methodology and presents a proposal with practical applications in agronomy. The combination of YOLOv10n-SSE with hybrid planning like APF + RRT is a novel contribution to agricultural robotics, particularly for crops with high occlusion. This system could reduce harvesting costs and labor dependency, aligning with sustainable agriculture objectives.  

Regarding visión, Although variable conditions such as lighting and occlusion are mentioned, the number of images corresponding to each scenario is not specified. The description includes images with pineapples of different sizes and varying levels of obstruction due to plant leaves. A quantitative breakdown is recommended to assess generalization.  

To evaluate the model’s effectiveness, a comparison with other recent models not based on YOLO (such as DETR or Faster R-CNN) is missing.  

The compensation applied to 2D path nodes for comparison with 3D is confusing. While it appears effective, it might introduce bias. A clearer explanation is needed.  

Another potential issue is that the hybrid algorithm assumes an ideal X-Z trajectory before planning in X-Y. However, It would be valuable to include a table outlining movement constraints and operational limitations of the robot, discussing challenging scenarios and justifying why certain cases may not be relevant.  

The article does not address how extreme cases such as pineapples partially outside the camera frame or with significant oclusión are handled. Additionally, the impact of depth estimation errors is not discussed. The authors should expand the discussion to include potential failure cases of the system.  

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors took all my comments into account. However, there are certain points that need clarification.

1.The authors noted that they added a figure (a comprehensive flowchart of the AIR algorithm ) on page 7, but it is not present in the revised version. Please check this.

2. In response to one of my comments, the authors stated as follows “Excellent suggestions! We have not only revised Table 2 but also added standard deviation values to Table 3.” But this is not in the corrected article (the standard deviation data is missing from these tables). Please check this.

I have no other comments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf