Swiss Round Selection Algorithm for Multi-Robot Task Scheduling
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis paper addresses the topical problem of multi-task allocation of a group of robots, which is particularly relevant for picking and packing tasks in e-commerce centres. A Swiss round selection algorithm is proposed to solve this problem and its superiority over Tournament Selection, Elite Selection and Roulette Wheel Selection algorithms is verified by simulations. It is observed that the superiority of the proposed algorithm is highest with 3 robots and decreases with increasing number of robots. With 6 robots, the simulation results are the same for all algorithms.
Several inaccuracies were observed in the paper.
Does Table 1 Definition S correspond to the Object Phase?
Selection Operation in Figure 7 does not correspond to the description in Step 3.
Conclusions should be improved.
In addition, the orthogonal trajectory on the grid and the return to the initial position are not optimal, and the algorithm does not explore the possibility of employment of the robot. These and the authors' chosen limitations undermine the applicability of the research results in real-life settings.
Comments on the Quality of English LanguageNo comments.
Author Response
Dear reviewers:
Thank you for your careful review of the paper. Your review comments are very important to us and have helped us to revise and improve the content of the paper. It is also an important contribution to the further development of this research field. We have revised and improved the paper and resubmitted the manuscript. Please see the revised manuscript for the specific revisions. The following is our point-by-point response to your comments:
Several inaccuracies were observed in the paper.
Does Table 1 Definition S correspond to the Object Phase?
Response: In Table 1, the object corresponding to S is defined as the shelf, and this has been corrected in the revised paper.
Selection Operation in Figure 7 does not correspond to the description in Step 3.
Response: In Figure 7, the selection operation and description were indeed inconsistent. The numbers corresponding to Group D, Group C, and Group A after the selection in step 3 of the original figure have been revised. Group D should be 30, Group C should be 90, and Group A should be 30.
Conclusions should be improved.
Response:We have improved the conclusion section, providing a more accurate description of the characteristics of the algorithm presented in this paper and the results achieved. Please refer to the conclusion section of the revised paper for details.
In addition, the orthogonal trajectory on the grid and the return to the initial position are not optimal, and the algorithm does not explore the possibility of employment of the robot. These and the authors' chosen limitations undermine the applicability of the research results in real-life settings.
Response: Thank you for your feedback. At the outset of this paper, we considered the proximity of shelves to each other in actual warehousing environments, which is why we only considered the robots' movement in four directions. Additionally, the focus of our research was on the performance of the genetic algorithm, and through experiments, we have demonstrated that when using the Manhattan distance (where robots can only move in four directions), the Swiss round selection algorithm proposed in this paper is superior to the genetic algorithm.
The suggestions you have made are also directions for our future research. After conducting multi-robot scheduling, we will consider the path planning issues during robot movement, allowing robots to move in multiple directions and find the optimal path, thereby enhancing applicability in practical environments. We will further investigate these issues in our subsequent research.
In terms of language expression, this paper uses the editing service provided by Editsprings to ensure the accuracy and professionalism of the paper's language expression and grammar. The editing certificate can be verified on the website https://www.editsprings.com/QueryCertificate.html. Enter the certificate number "ES-202404231557817751" on the above website to verify the certificate.
If you have any questions, feel free to contact us at any time. Have a great day!
Best Regards，
Zucheng Huang
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper does not contain a link do the software and datasets used in the experiments, what is a fundamental failure and papers without software and datasets should never be accepted for publications, as there is no way to recreate the experiments and verify the authors' method. First of all, the authors must provide a link do the software and datasets used in the experiments.
Other remarks:
The "crossover" operation (Fig. 5) is rather a 2-point mutation operation, as the child is composed from only one parent.
The 1-point and 2-points mutations (the latest one called "crossover" by the authors seems to provide small changes, especially in longer chromosomes, so I doubt the method will be effective.
The following fragment is totally unclear:
"This paper does not set a crossover probability, and all parent chromosomes are involved in the crossover operation. This is because the two-phase tasks of the e-commerce logistics and warehouse have time constraints, and through this strategy, we aim to reduce the time overhead of the crossover operation. The specific process of the crossover operation is as follows:"
I cannot find the explanations of how are the two kind of chromosomes evaluated. Since the task encoding chromosome is not being modified in the optimization process, what is the purpose of using it?
The selection is so complex? It is as matter of fact based on a sequence of tournament selections. Why did you design the selection algorithm in this way? What was the rationale behind that design?
What do you understand by "full crossover and full mutation operations"?
Author Response
Dear reviewers:
Thank you for your careful review of the paper. Your review comments are very important to us and have helped us to revise and improve the content of the paper. It is also an important contribution to the further development of this research field. We have revised and improved the paper and resubmitted the manuscript. Please see the revised manuscript for the specific revisions. The following is our point-by-point response to your comments:
The paper does not contain a link do the software and datasets used in the experiments, what is a fundamental failure and papers without software and datasets should never be accepted for publications, as there is no way to recreate the experiments and verify the authors' method. First of all, the authors must provide a link do the software and datasets used in the experiments.
Response: This paper does not involve a dataset. We have already uploaded the original code to Gitee, and the link is: https://gitee.com/liii458/swiss-choice-algorithm.git
Other remarks:
The "crossover" operation (Fig. 5) is rather a 2-point mutation operation, as the child is composed from only one parent.
The 1-point and 2-points mutations (the latest one called "crossover" by the authors seems to provide small changes, especially in longer chromosomes, so I doubt the method will be effective.
Response: The two issues mentioned above are both related to the crossover operation. In the crossover operation, the method used in this paper involves crossing over the genes in the robot encoding of the parent chromosomes. After the crossover operation, the corresponding tasks performed by the robots are changed. Since the tasks in the warehousing model of this paper are divided into two stages, the first stage tasks must be completed before the second stage tasks; otherwise, the robots executing the second stage tasks will have to wait. Therefore, after the crossover operation in this paper, the order of task execution will be changed, which will effectively affect the scheduling time of the entire system. Additionally, because all parent chromosomes participate in the crossover operation in this paper, the range that the population can search is also very large, which will make the algorithm faster to reach the optimal solution.
The following fragment is totally unclear:
"This paper does not set a crossover probability, and all parent chromosomes are involved in the crossover operation. This is because the two-phase tasks of the e-commerce logistics and warehouse have time constraints, and through this strategy, we aim to reduce the time overhead of the crossover operation. The specific process of the crossover operation is as follows:"
I cannot find the explanations of how are the two kind of chromosomes evaluated. Since the task encoding chromosome is not being modified in the optimization process, what is the purpose of using it?
Response: This paper employs a dual-chromosome encoding method because the robot numbering needs to correspond one-to-one with the task encoding. In the warehousing model used in this paper, tasks are divided into two stages. The first appearance of the task encoding corresponds to the robot number executing the first-stage task, and the second appearance of the task encoding corresponds to the respective robot executing the corresponding second-stage task. During the algorithm's crossover and mutation process, although the task encoding is not modified, the execution order of the corresponding tasks will change as the robot encoding is altered. Essentially, the sequence in which tasks are performed is also altered, which is beneficial for the algorithm to find the optimal scheduling solution.
The selection is so complex? It is as matter of fact based on a sequence of tournament selections. Why did you design the selection algorithm in this way? What was the rationale behind that design?
Response: The Swiss-system tournament format originated in the 1985 chess competition held in Zurich, Switzerland, and is now widely used in various competitive events. In the original genetic algorithm, due to its selection method being essentially a selection of the best, this can lead to the algorithm possibly falling into a local optimum, which is not conducive to the development of population diversity. Therefore, I proposed a Swiss round selection mechanism with an upset probability. The purpose of the design is to ensure that high-quality individuals are smoothly inherited, while also allowing some medium and poor individuals to be inherited to the next generation. In the selection process, the highest quality individuals will directly enter the next generation after being screened. When other individuals are compared using the fitness function, a certain upset probability is introduced to ensure that medium and poor individuals have the opportunity to enter the next generation. This not only retains the genes of high-quality chromosomes but also enhances the diversity of the population, effectively preventing the algorithm from falling into a local optimum.
What do you understand by "full crossover and full mutation operations"?
Response: Complete crossover and mutation operations mean that all parent chromosomes undergo crossover and mutation operations, unlike the original genetic algorithm where crossover and mutation probabilities are set. In the original genetic algorithm, parent chromosomes after crossover and mutation are not retained, which is why crossover and mutation probabilities are set to protect high-quality parents from participating in crossover and mutation operations, thus ensuring the smooth inheritance of high-quality parental genes to the next generation.
In the algorithm presented in this paper, because the original parent chromosome genes before crossover and mutation are retained after the crossover and mutation, and are placed together with the newly generated offspring chromosomes to participate in the new round of selection operations, high-quality parents and offspring chromosomes will be successfully retained to the next generation in the selection process. Therefore, there is no need to set crossover and mutation probabilities to protect high-quality parental chromosomes, so complete crossover and complete mutation operations can be adopted.
This approach expands the search range of the population, increases the diversity of the population, ensures that the population can expand in multiple directions, effectively avoids the algorithm falling into local optima, and significantly improves the efficiency of the algorithm.
If you have any questions, feel free to contact us at any time. Have a great day!
Best Regards，
Zucheng Huang
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors' explanations are good, but the link to the source code was placed only in the authors' answer to the reviewer (to me), but not in the paper. Please place the link in the paper also. This is currently my only comment.
Author Response
Dear reviewer,
Thank you for your timely response, which are very helpful for us to further refine the paper. Based on your comments, we have put the data link in the article, we have resubmitted the manuscript, please review the data availability section. Thank you again for your valuable comments and have a great day.
Best Regards，
Zucheng Huang
