An Off-Line Error Compensation Method for Absolute Positioning Accuracy of Industrial Robots Based on Differential Evolution and Deep Belief Networks

Round 1

Reviewer 1 Report

Manuscript ID: electronics-2573517

Journal: Electronics

Title: An off-line error compensation method for absolute positioning accuracy of industrial

                                 robots based on differential evolution and deep belief networks

Authors: Yong Tao, Haitao Liu, Shuo Chen, Jiangbo Lan, Qi Qi and Wenlei Xiao

Intelligent manufacturing has seen an increase in the use of industrial robots.  One of the challenges in applying industrial robots is their poor absolute positional precision. Based on deep belief networks and an off-line compensation technique, an absolute positioning accuracy compensation algorithm for industrial robots is developed. To enhance the networks, a differential evolution algorithm is provided. A position error map ping model is suggested in conjunction with the evidence theory to achieve the absolute positioning accuracy compensation of industrial robots. The robot's end's absolute position error is decreased from 0.469 mm to 0.084 mm, with an increase in accuracy of 82.14%. According to experimental findings, the suggested compensation mechanism has the potential to increase the absolute positioning accuracy of industrial robots and be useful for a variety of precise operational tasks.  The paper is well written, interested and the results are good, I would like to suggest the following MINOR corrections before acceptance:

ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ

(1)  A professional proofreading revision is strongly required. Typos must be corrected.

(2)  Please add more details about the studied model

(3)  The introduction must be reformulated to contain literature and future works, the main aim of the work. Also, the arrangement of the manuscript should be added in a paragraph at the end of the introduction.

(4)  The authors should state clearly in the introduction the advantages of the used technique and a summary of the literature.

(5)  The results are interesting, also there was a great effort done, but if the authors can add some applications of their results this will be great.

(6)  I didn’t see any mathematical model for the problem under study, I think if there is a mathematical model that describes the studied problem it should be added or minimum referring to it in some references.

(7)  Some new works on the studied problem should be added, this will improve the paper.

(8)  The authors should revise and carefully arrange the references according to the guidelines of the journal.

Thanks a lot, to the editorial board of the Electronics  Journal.

Comments for author File: Comments.pdf

Author Response

Dear editor

Thank you for your comments concerning our manuscript entitled “An off-line error compensation method for absolute positioning accuracy of industrial robots based on differential evolution and deep belief networks”. Those comments are all valuable and helpful for revising and improving our paper, as well as the important guilding significance to our researches. We have stutied comments carefully and have made correction which we hope meet with approval. Revised Portion are marked in red in the paper. The main corrections in the paper and responds to the review’s comments are as following:

• A professional proofreading revision is strongly required. Typos must be corrected.

Response: We are very sorry for our Typos. We have proofread it,and Typos has been modified

• Please add more details about the studied model

Response: Thanks for your comments, some details have been added in the article. Above Figure 4, We add an explanation of the relationship between the fitness function of DE algorithm and DBN. Above Section 2.2, We add that DBN is good at handling the nonlinear problem of error compensation because of strong robustness and fault tolerance.

• The introduction must be reformulated to contain literature and future works, the main aim of the work. Also, the arrangement of the manuscript should be added in a paragraph at the end of the introduction.

Response: Thank you very much for your advice. The introduction has been reformulated, which contains literature that other researchers worked on and future works. The arrangement of the manuscript has been added in a paragraph at the end of the introduction.

• The authors should state clearly in the introduction the advantages of the used technique and a summary of the literature.

Response: The technique used in this paper to improve the absolute positioning accuracy of robots (based on DBN and DE) have been added in the paper. The added part is the part in red on the top of Figure 1. At the same time, the advantages of DBN, DE and evidence theory are introduced

• The results are interesting, also there was a great effort done, but if the authors can add some applications of their results this will be great.

Response: Thank you for your recognition, we will continue to work hard. Your comments are very reasonable, and the application of the results of this paper has been added to the paper.

• I didn’t see any mathematical model for the problem under study, I think if there is a mathematical model that describes the studied problem it should be added or minimum referring to it in some references.

Response: We think what you say makes sense. Formula (1)-(8) is the mathematical model of neural network, including the establishment of DBN, the selection of loss function. Formula (9)-(13) is the mathematical model of the differential evolution algorithm. Formulas (14)-(18) are mathematical models of evidence theory used to make experimental results more convincing.

• Some new works on the studied problem should be added, this will improve the paper.

Response: Thanks for your comments, some new works on the studied problem have been added. There are many mathematical methods [29–32] that can also be used to study the positioning accuracy of robots. Analytic methods such as fractional order have gained more and more attention [33,34].

• The authors should revise and carefully arrange the references according to the guidelines of the journal.

Response: Thank you for your helpful advice, we have revised and carefully arranged the references according to the guidelines of the journal.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper shows off-line error compensation method for the position accuracy of industrial robots. This work is helpful for industry and I cannot find any specific technical issues in this paper.

1) Similar work can be compared. What is the original and meaningful work compared to the previous papers. you can cite the more relavant papers.

2) Red line should be removed on EtherCAT in Figure 8.

3) The definition of the abbreviation must first be well defined.

4) Position is same locaterion for position error in Figure 18?

NA

Author Response

Dear editor，

Thank you for your comments concerning our manuscript entitled “An off-line error compensation method for absolute positioning accuracy of industrial robots based on differential evolution and deep belief networks”. Those comments are all valuable and helpful for revising and improving our paper, as well as the important guilding significance to our researches. We have stutied comments carefully and have made correction which we hope meet with approval. Revised Portion are marked in red in the paper. The main corrections in the paper and responds to the review’s comments are as following:

• Similar work can be compared. What is the original and meaningful work compared to the previous papers. you can cite the more relavant papers.

Response: We quite agree with your suggestion and have added more references to the article. References 29-34 are an analysis of some mathematical methods. References 35-38 are the original and meaningful work compared to the previous papers.

• Red line should be removed on EtherCAT in Figure 8.

Response: Thanks for your comments, and red line have been removed on EtherCAT in Figure 8

• The definition of the abbreviation must first be well defined.

Response: Thank you very much for your advice. We have checked the paper to assure that the definition of the abbreviation must first be well defined.

• Position is same locaterion for position error in Figure 18?

Response: After off-line compensation, the minimum value is reduced from 0.097 mm to 0.006 mm. The average value is reduced from 0.110 mm to 0.083 mm. The position error after compensation and the previous position will remain in the trend.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors present a method to compensate the positional accuracy of the end-effector of an industrial KUKA robot. The proposed method is the DE-DBN which is well described. The work is technically sound, however, there are some aspects that should be considered to improve the readability of the manuscript.

From the robot datasheet is evident the repeatability value which is +/- 0.3mm. However, positional accuracy is not accounted for due to the high dependency on the payload, arm configuration, stiffness, temperature, etc. This is why many manufacturers do not include accuracy in the robot’s datasheet.  In the article you established an absolute positional accuracy of +/- 0.6 mm, please explain under which circumstances this accuracy was obtained and if this accuracy is repeatable.

Using DBN for error compensation can be computationally expensive. This is especially important when dealing with robot operations in continuous mode during real-world operations.  Is it appropriate your method? What is the convergence time before doing a step robot’s motion. Please define the scope and limitations of your research and mention if it is possible to implement your method in real-time? What are the considerations?

typo:

Line 155. “is iteration rate” is ambiguous please amend.

Author Response

Dear editor，

Thank you for your comments concerning our manuscript entitled “An off-line error compensation method for absolute positioning accuracy of industrial robots based on differential evolution and deep belief networks”. Those comments are all valuable and helpful for revising and improving our paper, as well as the important guilding significance to our researches. We have stutied comments carefully and have made correction which we hope meet with approval. Revised Portion are marked in red in the paper. The main corrections in the paper and responds to the review’s comments are as following:

• From the robot datasheet is evident the repeatability value which is +/- 0.3mm. However, positional accuracy is not accounted for due to the high dependency on the payload, arm configuration, stiffness, temperature, etc. This is why many manufacturers do not include accuracy in the robot’s datasheet. In the article you established an absolute positional accuracy of +/- 0.6 mm, please explain under which circumstances this accuracy was obtained and if this accuracy is repeatable.

Response: Thank you for your comments. The absolute positioning accuracy of the robot is obtained from the KUKA website

(https://xpert.kuka.com/service-express/portal/project1_p/document/kuka-project1_p-basic_AR6695_en?context=%7B%22filter%22%3A%7B%7D,%22text%22%3A%22kr%206%20r700%20sixx%20CR%22,%22useExpertQuery%22%3A0%7D). And the robot, the laser tracker and the environment have some influence on the experimental results. The results of the experiment can be repeated within the error tolerance.

• Using DBN for error compensation can be computationally expensive. This is especially important when dealing with robot operations in continuous mode during real-world operations. Is it appropriate your method? What is the convergence time before doing a step robot’s motion. Please define the scope and limitations of your research and mention if it is possible to implement your method in real-time? What are the considerations?

Response: Thank you very much for your advice. Our approach is an off-line compensation technique. This method trains the model offline and then deploits it to the robot. So our method is appropriate for dealing with robot operations in continuous mode during real-world operations.  There is no convergence time due to offline compensation.Our research scope is the off-line compensation method for absolute positioning accuracy of industrial robots. The limitation of this method is that the model needs to be trained before being deployed to the robot, not online. This modification has been marked in red in the paper. Considerations include an experimental environment free of vibration, the allowable operating temperature of the robot, and the higher accuracy of the laser tracker. This modification has been marked in red in the paper at lines 529-531.

• typo: Line 155. “is iteration rate” is ambiguous please amend.

Response: Thanks for your comments, and “ is iteration rate” has been modified to “ is iteration rate of the DE algorithm”. This modification has been marked in red in the paper at line 270.

Author Response File: Author Response.pdf

Reviewer 4 Report

The methodological and results sections are clear strengths of the scholarship.

The areas needing great improvement are the Introduction and Conclusions respective to Industry 4.0 technologies and there merits, diversity, and potential scientific and practice impact on actors. The following feedback below is an attempt to assist authors improve the first paragraphs of the Introduction and then to allow the authors to revise the Conclusions with a more holistic representation of the impact robotics can make, regardless of industry, to improve societal conditions given global population demands, effects from climate change, supply chain issues, health, etc. These additions would enhance not only the scientific merits of the scholarship by improving the document’s ability to increase translational research but offer global readers a sense of the author’s science improving practice not just science. Further, these inclusions would improve the scholarship’s ability to produce sustainable readership and citations beyond the potential publication of the manuscript. Introducing Industry 4.0 technologies is a must before discussing the literature/need to examine the impact from robotics implementations. 

1. Introduction; first paragraphs – for example…

Industry 4.0 technologies are critical and indispensable tools to propel social and technological innovation advancements https://doi.org/10.1016/j.techfore.2020.120332.

https://doi.org/10.3390/app10186498 described the use of Industry 4.0 technologies can better sustain current resources, reduce labor costs, can be better sources of energy, and potentially produce higher quality sustainable products. Examples of Industry 4.0 technologies include, but are not limited to, machine learning, virtual and augmented reality, IoT, artificial intelligence, big data, and robotics https://doi.org/10.1016/j.compind.2020.103300.

https://doi.org/10.1080/00207543.2019.1672902 found Industry 4.0 technologies assist manufacturing companies sustainability and increasing their economic potential. Scholars have examined Industry 4.0 technologies in diverse industries besides manufacturing. https://doi.org/10.1016/j.health.2021.100008 implemented a systematic review to understand the Industry 4.0 technologies use in managing the pandemic. Industry 4.0 technologies uses are ubiquitous to meet the increasing demands of society such as robotics https://doi.org/10.1016/j.gfs.2019.100347,  artificial intelligence https://doi.org/10.1016/j.aac.2022.10.001, IoT https://doi.org/10.3390/s22186833,  augmented reality https://doi.org/10.1016/j.techsoc.2021.101739, big data https://doi.org/10.1016/j.agsy.2021.103298, and machine learning https://doi.org/10.1080/10942912.2022.2066124 in food and agricultural sciences to assist the more efficient and enhanced production needed to feed a growing world populous. https://doi.org/10.1080/00207543.2020.1824085 investigated the use of Industry 4.0 technologies in the manufacturing sector in 380 papers prior to 2020. https://doi.org/10.1016/j.ijpe.2019.01.004 sought to understand the manufacturing patterns implemented from Industry 4.0 technologies. Robotics is a significant Industry 4.0 innovation that offers immeasurable possibilities in manufacturing disciplines https://doi.org/10.1016/j.cogr.2021.06.001.

2. Then on to the authors existing first paragraph which would be there second or third paragraph due to the new additions.

3. The above strengthening of the Introduction will warrant the revision and rebuilding of the Conclusions section and to better provide readers how and to what extent the author's scholarship improves both science and practice. 

Author Response

Dear editor,

Thank you for your comments concerning our manuscript entitled “An off-line error compensation method for absolute positioning accuracy of industrial robots based on differential evolution and deep belief networks”. Those comments are all valuable and helpful for revising and improving our paper, as well as the important guilding significance to our researches. We have stutied comments carefully and have made correction which we hope meet with approval. Revised Portion are marked in red in the paper. The main corrections in the paper and responds to the review’s comments are as following:

• Introduction; first paragraphs.

Response: Thank you very much for your advice. Your suggestion is very good, and industry 4.0 as the fourth industrial revolution, is a big trend that needs to be considered comprehensively. As an important participant in the automated manufacturing process, industrial robots directly affect the level of industrial manufacturing automation. We have made a supplement as the first  paragraph according to your comments.

• Then on to the authors existing first paragraph which would be there second or third paragraph due to the new additions

Response: Thank you for reminding. The existing first paragraph have become the second paragraph.

• The above strengthening of the Introduction will warrant the revision and rebuilding of the Conclusions section and to better provide readers how and to what extent the author's scholarship improves both science and practice

Response: Thanks for your comments. We think your suggestion is very reasonable. We have revised and rebuilded the Conclusions section. We summarize the proposed methods, experimental equipment, final effects, and future work.

Author Response File: Author Response.pdf

Reviewer 5 Report

The article proposes a compensation algorithm to improve the absolute positioning accuracy of industrial robots using deep belief networks, differential evolution, and a position error mapping model combined with evidence theory. The experiments were conducted on an industrial robot KR6_R700 using a laser tracker AT901-B. The results showed that the proposed compensation method significantly improved the absolute positioning accuracy of industrial robots, reducing the absolute position error of the end of the robot from 0.469 mm to 0.084 mm with 82.14% accuracy improving after the compensation. The study concludes that the proposed compensation method has the potential to improve the absolute positioning accuracy of industrial robots and can be used for precise operational tasks.

From my point of view, the article can be accepted in the present form.  

Small writing errors was observed (e.g. line 214: 3. supervised --> Supervised)

Author Response

Dear editor,

Thank you for your comments concerning our manuscript entitled “An off-line error compensation method for absolute positioning accuracy of industrial robots based on differential evolution and deep belief networks”. Those comments are all valuable and helpful for revising and improving our paper, as well as the important guilding significance to our researches. We have stutied comments carefully and have made correction which we hope meet with approval. Revised Portion are marked in red in the paper. The main corrections in the paper and responds to the review’s comments are as following:

• Small writing errors was observed (e.g. line 214: 3. supervised --> Supervised)

Response: We are very sorry for the writing errors. We have modified it and checked the full paper.

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

The authors have greatly enhanced this iteration of the scholarship. I hope the authors find the manuscript sustains its readership over a period of time.