Apple-Picking Robot Picking Path Planning Algorithm Based on Improved PSO
Round 1
Reviewer 1 Report
The authors propose an apple picking robot picking path planning algorithm based on improved PSO that dynamically adjust the velocity weights based on the trend of the particle fitness value and the position of the particle swarm center of mass. The authors evaluated their proposed algorithm and compared it with the standard PSO algorithm experimentally. The aim of their work is to reduce the collision between the manipulator and branches during apple picking and to improve the picking success rate and picking efficiency. The quality of the paper could be improved and the authors should take the following comments into consideration:
1- Section 1 (Introduction) should be divided into main sections: Introduction and Related Work. In the related work, the authors should present a deep analysis of the related work by introducing a table that compare between the references discussed in the related work.
2- The phrasing of some statements need reconsideration. As an example, the following statement was presented in the Abstract (“… is proposed, and the algorithm is used to The algorithm is used to plan..). Furthermore, some statements are too long.
3- Did authors test other algorithms other than RANSAC to eliminate the abnormal data in the point cloud that could be more efficient than the used algorithm? Please explain. are there any results that could be presented and indicate the efficiency of this algorithm?
4- In Table 1, is the apple picking cycle includes the time for calculating the apple picking direction or it just involves the time to run the PSO algorithms.
5- Are the results presented in Figure 7 for the improved PSO? If this is the case, how the figure would be with the standard PSO? Figure 7 was not cited clearly in the document. It seems that the first paragraph in page 13 explains Figure 7. Is that true? Please explain.
6- In the Discussion section, the authors pointed out to the effect of the closeness of the branches and they aim to study that effect in the future work. However, for the current model, are there any results that could be presented to show the effect of the branches closeness to the failure and success rate?
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper presents a novel PSO algorithm for an apple-picking robot path planning problem.
Overall, the paper is good and publishable. The authors need to go over extensive English editing.
- Citations need to be corrected. Example: use either et al. or numbers [1].
- the algorithm shows clear improvement over the standard PSO algorithm
- Some points on future work should be mentioned.
- The algorithm shows a high success rate compared to existing PSO. However, is that only because of the new PSO or other additions such as Yolo?
- Does the algorithm still works if the terrain is changing in real time? Please comment on the assumptions.
- Can the end effection poke and destroy the fruit in cases where the fruit doesn't fit the spherical assumption?
- Are there any metrics on how many fruits are picked but damaged?
Overall, the paper is good. It would be nice if the authors could answer the above questions.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Although the subject of the article does not seem to be best suited with the scope of the journal, the paper is relatively well written. Some comments and suggestions that could be addressed in the final version of the paper are given below.
1. How many apple trees are taken into account within the experiment? How many apples (counted "manually") were available on these trees? The authors say that "(...) 158 fruits extracted by the vision algorithm were picked.". This suggests that only apples identified by the vision algorithm were considered. Were there any apples in the tree that the algorithm couldn't identify? In order to properly assess the effectiveness of the algorithm, it would probably be necessary to take into account a larger number of apples and a larger number of different trees.
2. Does the proposed algorithm have any limitations regarding tree size etc? The question is not about the limitations of the mechanical part of the robot, but about the limitations of the algorithm, vision system, etc.
3. How long did it take for the robot to collect these 158 apples? How would you compare this time with the time needed for a person collecting the same amount of apples?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Minor spell check is required.
Author Response
Thank you very much for your suggestion, we have checked and revised the vocabulary in the text and adjusted some grammatical problems. Please let us know if there are any other problems, thank you.
