SA-SVM-Based Locomotion Pattern Recognition for Exoskeleton Robot

Round 1

Reviewer 1 Report

• Abstract is replete with unnecessary acronyms detracting from the manuscript’s attractiveness to readers. Particularly in typical manuscripts filled with mathematically based developments, the readers strongly rely on the text verbiage to elaborate and tie together the equations, the figures and tables of data. Consider a typical reader who scans the abstract to gauge personal interest in reading further.  The reader’s eyes land on line 28 (one example) first, “…SVM….Max-Relevance…Min-Redundancy…”.  Please replace acronyms and capitalized, special names as much as possible in the abstract seeking to keep it in broadest possible terminology.  If well-written, the reader will be enticed to continue to the introduction where such non-general terminology is appropriate. One mere example is offered:
  • g. In line 22, “a Simulated Annealing (SA) based Support Vector Machine (SVM) model” is completely meaningless to readers not already knowledgeable in such matters. Especially since the sentence articulates a proposed development, comprehension by the reader is paramount, and use of such nuanced verbiage will dissuade much of the readership.
• Use of accuracy percentages in the Abstract is a particularly strong technique.
• Immediately following accuracy percentages in line 35 is a sentence beginning, “This method…”. Which method is meant?  The experiments?  The paragraph reads that way.
• Acronyms in the keywords? SVM, but not SA?  FSM (fast steering mirrors)?  Readers scanning the abstract and keywords seeking to ascertain personal interest are too likely to put this manuscript down after reading the abstract and key words having not read further.
• Literature review is very well done with elaboration of rationales for investigating nearly every individual, cited reference.
• Figure 1 and 2 are very good and significantly enhance the manuscript’s transmission of the authors’ meanings and intent.
• Figure 3 is useless, since it relays no information due to complete illegibility. If the figure was completely omitted the manuscript would lose nothing. The figure must be improved with a reminder the smallest text size permissible in the manuscript template is the figure caption (a convenient place to gauge legibility while drafting the manuscript).
• Particularly due to the commonplace practice of potential readers initially scanning manuscripts (emphasizing figures and tables), the quality of figure 4 should be improved to reduce blurriness and normalize the internal text sizes to the rest of the document. In order to transmit meaning better, avoid acronyms in the figure captions.
• Figure 5 is illegible and relays no information, thus could be left out of the manuscript without changing its quality. This fact is amplified when considering reading the manuscript in printed hard copy (often on black and white printers).
• Figure 7 is illegible and relays no information, thus could be left out of the manuscript without changing its quality. This fact is amplified when considering reading the manuscript in printed hard copy (often on black and white printers). The caption does very little to aid the understanding of the figure’s content.
• Figure 8 is illegible and relays no information, thus could be left out of the manuscript without changing its quality. This fact is amplified when considering reading the manuscript in printed hard copy (often on black and white printers).
• Figure 9 is illegible and relays no information, thus could be left out of the manuscript without changing its quality. This fact is amplified when considering reading the manuscript in printed hard copy (often on black and white printers). The step-wise procedure for mRMR algorithm is more appropriately displayed as a table rather then in-line text to amplify the ease of attempts to repeat the research.
• Equations and tables are well done and enhance the manuscript’s quality and readability.
• Heading 2.3 and Heading 2.3.2 is poorly worded, requiring the reader to flip back through the pages to remind themselves of acronym definition before understanding the upcoming paragraph’s intended content. Meanwhile, heading 2.3.1 is well done.
• Tables 5, 6, and 7 are excellently presented.
• Use of accuracy percentages in the Discussion is particularly strong.
• The conclusions are very short and recommendations for future research are completely omitted, denying the reader the chance to place to proposal into the lineage of research. Particularly due to high flexibility, the reviewer would like to see the applicability of the recently published, so-called whiplash compensation for flexible robotics.

Author Response

Dear editor and dear reviewer,

Re: Manuscript ID: applsci-1221222 and Title: SA-SVM-Based Locomotion Pattern Recognition for Exoskeleton Robot

 

Thanks for your letter and for the Reviewer’s comments concerning our manuscript. Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in yellow in the paper. The main corrections in the paper and the responds to the reviewer’s comments are as following:

Responds to reviewer’s comments:

1. Comment: Abstract, key words and subtitles are replete with unnecessary acronyms.

         Response: Your advice is a great help to me. I have corrected unnecessary           abbreviations in the article.

2. Comment: Immediately following accuracy percentages in line 35 is a sentence beginning, “This method…”. Which method is meant? The experiments? The paragraph reads that way.

         Response: Your advice is very important. As I have written the colloquial             description into the paper, there are indeed some ambiguities in the                     article. I have revised the ambiguities in the full text.

3. Comment: Readers scanning the abstract and keywords seeking to ascertain personal interest are too likely to put this manuscript down after reading the abstract and key words having not read further.

        Response: Your suggestion is right. I ignored the reader's feelings when I            write the article. I have modified the key words.

4. Comment: Figures relay no information due to complete illegibility.

         Response: I am very sorry that I did not notice that the pictures were                   compressed in the process of saving, which made the pictures illegible                 and  affected the reviewer's review. I have replaced all the blurry images.

5. Comment: The conclusions are very short and recommendations for future research are completely omitted.

        Response: Your suggestions are of great help to me, and the other two                reviewers also put forward some suggestions for my conclusion. I rewrote            the discussion and conclusion section of the article. A discussion of model          defects and a summary of model performance are added in this paper.

Combined with the suggestions of the other two reviewers, I also revised some grammatical and formatting issues in the article, and added relevant documents. I also added the state transition diagram of the finite state machine and the flow chart of the mRMR algorithm.

Special thanks to you for your good comments. We are extremely grateful to Reviewer for pointing out this problem. Thank you and all the reviewers for the kind advice again. If you have any questions, please contact us without hesitate.

Yours sincerely

Zeyu Yin

Author Response File: Author Response.docx

Reviewer 2 Report

Identifying walking patterns is an exciting research approach. However, I believe that the study is missing many elements to make it useful for the research community.

The contribution of the work is not clear, and the novelty is missing. Different types of walking should be included, and sensory data need to be presented for each case.

 

This sentence is not correct and needs to be rephrased. “The main purpose of its design and manufacture is to help the wearer carry more heavy loads.”

It is not clear in Figure 8 where are the different types of walking.

The manuscript claims (in the last paragraph of the introduction) that there is an improvement in the system's real-time performance; however, this is not presented elsewhere in the manuscript!

Figure 4 and its related paragraph illustrate the working principle of SVM; I think it is not necessary to include them here since they are very common. My suggestion is to describe graphically the „SVM based on SA“.

Most of the figures appear in bad quality, which makes them unclear.

Figure 3 presents IMU data and FSR data, but the units are not presented. What are the horizontal and vertical axes and units? It is very unusual to see blurred figures in a manuscript. I wonder if the authors check the clarity of the figure.

Author Response

Dear editor and dear reviewer,

Re: Manuscript ID: applsci-1221222 and Title: SA-SVM-Based Locomotion Pattern Recognition for Exoskeleton Robot

 

Thanks for your letter and for the Reviewer’s comments concerning our manuscript. Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in yellow in the paper. The main corrections in the paper and the responds to the reviewer’s comments are as following:

Responds to reviewer’s comments:

1. Comment: Different types of walking should be included, and sensory data need to be presented for each case.

         Response: Your suggestion is of great help to my manuscript. In the                     revised manuscript, I modified the clarity of the picture and added the                  stairs ascent and stairs descent data pictures in Figure 3-5.

2. Comment: This sentence is not correct and needs to be rephrased. “The main purpose of its design and manufacture is to help the wearer carry more heavy loads.”

         Response: Your comments are correct. I've written this sentence too one-           sidedly. I changed the sentence to " The main purpose of its design and               manufacture is to enhance human strength, assist human walking and                 help patients recover".

3. Comment: It is not clear in Figure 8 where are the different types of walking.

         Response: Figure 8 in the original manuscript is too blurry, and I have                 adjusted the clarity of the picture. In the revised manuscript, the Y-axis                 coordinates in Figure 11-12 represent the different types of walking. 1                 represents level ground walking, 2 represents stairs ascent and 3                           represents stairs descent. Describe specifically on line 375-376. 1, 2, 3 in               the experiment are the labels of three locomotion patterns.

4. Comment: Figure 4 and its related paragraph illustrate the working principle of SVM; I think it is not necessary to include them here since they are very common. My suggestion is to describe graphically the “SVM based on SA”.

         Response: Your suggestion is of great benefit to my manuscript. In the                 revised manuscript, I deleted the picture of the SVM. In Figure 7, the                     process of parameter optimization of support vector machines by analog             annealing algorithm is introduced in detail.

5. Comment: Most of the figures appear in bad quality, which makes them unclear. Figure 3 presents IMU data and FSR data, but the units are not presented. What are the horizontal and vertical axes and units?

         Response: Your comments are very helpful to me. I ignored the reader's             feelings in the writing of the manuscript. I re-edited the Figure 3-5. I                     adjusted the clarity of the pictures and added explanations and units for             the Y and X axes.

Combined with the views of the other two reviewers. I added mRMR algorithm flowcharts and state transfer diagrams for finite state machines to the revised manuscript. I also rewoven the discussion and conclusion sections of the manuscript and modified the formatting and syntax errors of the article.

Special thanks to you for your good comments. We are extremely grateful to Reviewer for pointing out this problem. Thank you and all the reviewers for the kind advice again. If you have any questions, please contact us without hesitate.

Yours sincerely

Zeyu Yin

Author Response File: Author Response.docx

Reviewer 3 Report

Dear Authors.

Thanks for your paper. It introduces an interesting research.

Please find my comments bellow:

* Line 58. I do not understand why do you add this “ and stimulated by stimulation.” to the end of this phrase. Is it relevant that this electric potentials can be stimulated? Isn’t it more important that there are electric potentials than these potentials can be stimulated?

* Line 60. The proper acronym for Electroencephalography is “EEG” not “EMG”. Or maybe you wanted to write “Electromyography” instead of “Electroencephalography”....

* Line 62. Who is the experimenter? Do you refer to the person wearing the exoskeleton? Or maybe you meant “the experiment” instead of “the experimenter”....

* Line 64. I do not understand the sentence “the skin surface and intense movement,”. Maybe you meant “the skin surface under intense movement,” or “the skin surface at intense movement,”....

* Line 71. If you are not referring to a given human pattern recognition algorithm, but to a general way to HPR algorithms, then I suggest writing “Usually the human locomotion pattern recognition algorithms use a…”

* Line 73. Is there a reference to support this statement “However, Shao et al. “?

* Line 77. Add a “to” to “…Histogram represent features”. -> “…Histogram to represent features”.

* Lines 77  and 78. “KMP has demonstrated a leading position in human locomotion pattern recognition across five popular human locomotion data sets.”. Add reference to support this statement.
Can you please also reference the data sets?

* Line 115. “SA is one of the earliest and most popular meta-heuristic algorithms,”. The word "algorithm" should be singular because it refers only to “SA”.

* Lines 122 and 167. “Angle”. No capital letter.

* Line 123. The acronym “FSR” at “ joint angle information and FSRs data were combined” was not previously defined. Define it here please.

* Line 132. Add units to “frequency is 2G” -> “frequency is 2GHz”.

* Line 133. Use “whose” instead of “which”.

* Line 182. No capital letter at “of Filter.”.

* Line 223. I rather use a “.” instead of a “;” at “neon; The smaller t”

* Line 224. I think you should remove the “but” at “..ability, but the…”.

* Lines 226, 227 and 228. I suggest removing the three “is”

* Line 249- “Are” instead of “is”.

* Line 252. Do you mean equation number 11? Please indicate it as a reference.

* Lines 252 and 254. The verb is missing at sentence “the appropriate kernel function which can carry out nonlinear transformation in the input space”. Do you mean “the appropriate kernel function is the one that  can carry out nonlinear transformation in the input space”?

* Line 254. Capital letter at “space. it can…”

* Line 275. Include an “or at “…function, or Gaussian…” or add an “etc” at the end of the sentence “basis kernel 257 function (RBF), etc.”

* Line 258. At “RBF function is shown as follows” do you refer to equation 12? Please include a reference.

* Line 281. Something is missing at “This means that all the states have the same probability at high temperature. And as 281 the temperature goes down, there’s ???”

* Paragraph from line 312 to line 319. I suggest you include the description of the protocol you used to build the data set. How much time or how many meters the subjects walked? How many stairs did they climb or walk down? How many times did they repeat the protocol? Were they wearing a load or just walking with no extra load? etc.

* At lines 412 to 415 you say that your method is applied to the locomotion pattern recognition of the exoskeleton robot under different people, different terrain, different loads and different speeds. 
Do you refer to the moment when your system was used to build the data set? Or are you referring to two different situations: one when building the data set, and the other one when testing the final solution in real situations?

* Line 313 and 314. Better “The subjects are 160cm~180cm..” instead of “The subject is 160cm~180cm…”

* Line 321. ”data set” (singular) instead of “data sets”? Otherwise, in “It is divided into 70%” is should read “They are divided…”

* Line 325. What do you mean with “Through data preprocessing”? Through is not a verb, it is a preposition. Please clarify.

* Line 344. Substitute “.” by “,” and insert a blank space after the “,”.

* Line 351. What do you mean with “Locomotion mutation”? The change in the type of movement? I mean, changing quickly from walking to running? Please clarify. 

* Line 372. The title of the figure 9 is a widow paragraph. It should be linked to the figure. It should be in the same page as the figure.

* Line 386. Use “Tables” (plural) instead of “Table”.

* Line 406. Better “results” (plural) than “result”?

* Paragraph from line 367 to line 370. Please review the writing as it is difficult to understand. It is difficult for me to understand if each state of the FSM represent each locomotion pattern. It is also difficult for me to understand why qi+1, qi+2 and qi+3 are the same locomotion patterns as qi. If qi represent different states of the FSM, how can they be the same locomotion patter?

I suggest including a figure of the FSM where the states and transitions are easy to identify by the reader. I suggest using the international standard FSM modeling language called GRAFCET to describe your FSM.

* Lines 400 and 401. Please include references to support these statements.
Do that average accuracy of the Naive Bayesian method and NN method refer to the accuracy of these methods obtained when applied to “exoskeleton lower limb locomotion pattern recognition” or when applied to “locomotion pattern recognition” in general?
Please clarify. If Naive Bayesian and NN average accuracy refer to general “locomotion pattern recognition” then I am afraid they are not comparable to those obtained in your research.

* Lines 406 and 407. I suggest including deeper explanations in the “Results” section about the poor recognition results in locomotion modes transition, and about the speed of algorithm execution, and the delays in recognition. These issues are too relevant for the reader to understand the advantages and drawbacks of your suggested method, for just mentioning them with a single statement at the end of the paper.

Lines 409 to 412. Your method already include a FSM. Do you mean that you are going to modify that FSM to limit the transition between states? Please clarify.
 
* Lines 412 to 415. You do not say anything about performance when applying your method to those situations. I suggest you complete this sentence with some comments about performance.

* In my humble opinion, the “Results” section is mixing the presentation and explanation/description of results, with the description of some techniques used to obtain those results.
For example, the description of the “Experimental Scheme” and the description of the “FSM”, in my opinion, are not results, but part of the “Materials and methods” used in the experiment.
So, I suggest moving them into the “Materials and methods” section.

 

Finally, please be so kind as to revise all the paper for blank spaces before and after references, for example at line 51 “methods[1]” should be “methods [1]”.

* Also revise al the paper for blank spaces in general. The following are some examples:
* Line 35. “…accuracy.  Experiments…” It looks like if there is an extra blank space there.
* Line 51. Add blank spaces before and after the “,”. “..methods[1],biology…”
* Line 52. Add blank space “s [2-3]and…”
* Line 60. Add blank space “ (EMG)[5]and” between “[5]” and “and”
* Line 3. I think i better use “…to the approaches used to obtain information…” or “to the approaches of obtaining information”
* Line 76. Add blank space after “,” in “)[8] ,Local..”, and remove blank space after the “]” and before the “,”. 
* Line 76. Add blank space at “ (LBP)[9]and…”.
* Line 88. Add blank space before “T” at  “ is 91.6%.To “
* Line 114. Add blank space at “ SA-SVM.SA”
* Line 146. Add blank space at “ FSRs(LOSON LSH-10) “.
* Line 148. Add blank space at “ Figure 2.The m”.
* Line 157. Add blank space at “of 10Hz.Finally”.
* Line 211. Revise blank spaces at “problems[17].SVM “.
* Line 299. Revise blank spaces at “ (ACO)[20];Subasi e”.
* Line 300. Revise blank spaces at “ (PSO)[21];Wang “.
* Line 302. Revise blank spaces at “chines[22].Th”
* Line 304. Revise blank spaces at “ability.GA”
* Line 308. Revise blank spaces at Figure 5.SA-SVM model training“
 

Author Response

Dear editor and dear reviewer,

Re: Manuscript ID: applsci-1221222 and Title: SA-SVM-Based Locomotion Pattern Recognition for Exoskeleton Robot

 

Thanks for your letter and for the Reviewer’s comments concerning our manuscript. Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in yellow in the paper. The main corrections in the paper and the responds to the reviewer’s comments are as following:

Responds to reviewer’s comments:

1. Comment: I do not understand why do you add this “and stimulated by stimulation.”

         Response: This is my negligence. I did not pay attention to the accuracy             of the language in the process of translating the original text into English.           I've deleted “by stimulation”.

2. Comment: I do not understand why do you add this “and stimulated by stimulation.”

         Response: This is my negligence. I did not pay attention to the accuracy             of the language in the process of translating the original text into English.           I've deleted “by stimulation”.

3. Comment: Errors in acronyms, grammar, and formatting.

         Response: Your comments have been very helpful to my manuscript.                   I corrected the wrong location you pointed out article by article. I will keep           that in mind in the future. I have also made changes to sentences with an             unknown meaning.

4. Comment: “KMP has demonstrated a leading position in human locomotion pattern recognition across five popular human locomotion data sets.”. Add reference to support this statement.

         Response: I'm sorry I forgot to add this reference. In the revised                           manuscript, I re-added this reference [7] on line 73.

5. Comment: Something is missing at “This means that all the states have the same probability at high temperature. And as 281 the temperature goes down, there’s ???”

         Response: The meaning of this sentence is to draw the following formula.           I'm sorry to have caused you to be ambiguous. In the revised manuscript,             I deleted the sentence.

6. Comment: I suggest you include the description of the protocol you used to build the data set. How much time or how many meters the subjects walked? How many stairs did they climb or walk down? How many times did they repeat the protocol? Were they wearing a load or just walking with no extra load? etc.

         Response: I'm sorry I didn't write the protocol for building the dataset                 in the original manuscript. I add this section to the revised manuscript,                 lines 155-171.

7. Comment: At lines 412 to 415 you say that your method is applied to the locomotion pattern recognition of the exoskeleton robot under different people, different terrain, different loads and different speeds. Do you refer to the moment when your system was used to build the data set? Or are you referring to two different situations: one when building the data set, and the other one when testing the final solution in real situations?

         Response: I'm sorry I didn't make it clear in the original manuscript. These           four different applications are the acceptance criteria for our exoskeleton             robots. Our exoskeleton robots can eventually be used in these four                     scenarios. It is not strict for me to write this sentence into my manuscript.             In the revised manuscript, I deleted the sentence and rewrote the                         conclusion.

8. Comment: What do you mean with “Locomotion mutation”? The change in the type of movement? I mean, changing quickly from walking to running? Please clarify.

         Response: It was my negligence. I thought there was such a name in                   English. In lines 338-339 of the revised manuscript, I explained the                       locomotion mutation.

9. Comment: Please review the writing as it is difficult to understand. It is difficult for me to understand if each state of the FSM represent each locomotion pattern. It is also difficult for me to understand why qi+1, qi+2 and qi+3 are the same locomotion patterns as qi. If qi represent different states of the FSM, how can they be the same locomotion patter?

         I suggest including a figure of the FSM where the states and transitions               are easy to identify by the reader. I suggest using the international                       standard FSM modeling language called GRAFCET to describe your FSM.

         Response: Your suggestion is very important to my manuscript. The                     transfer conditions for the locomotion pattern are as follows: for the                     current locomotion pattern , determine whether , and  are the same                     locomotion pattern and the locomotion pattern is 's next locomotion                   pattern, if so,  moves to the next locomotion pattern. If not, the current               locomotion pattern remains unchanged. In the revised manuscript, I                     added the above explanation to the Finite State Machine State Transfer               map in lines 359-364.

10. Comment: Please include references to support these statements.

         Do that average accuracy of the Naive Bayesian method and NN method             refer to the accuracy of these methods obtained when applied to                         “exoskeleton lower limb locomotion pattern recognition” or when applied           to “locomotion pattern recognition” in general?

         Please clarify. If Naive Bayesian and NN average accuracy refer to general             “locomotion pattern recognition” then I am afraid they are not                             comparable to those obtained in your research.

         Response: Your comments are correct. I wanted to compare the proposed           algorithm horizontally with other algorithms, but I ignored the specific                 application scenario. In the revised manuscript, I deleted the sentence.

11. Comment: I suggest including deeper explanations in the “Results” section about the poor recognition results in locomotion modes transition, and about the speed of algorithm execution, and the delays in recognition. These issues are too relevant for the reader to understand the advantages and drawbacks of your suggested method, for just mentioning them with a single statement at the end of the paper.

         You do not say anything about performance when applying your method             to those situations. I suggest you complete this sentence with some                     comments about performance.

         Response: Thank you very much for your suggestion, which has helped               my manuscript a lot. The conclusion in the original manuscript I did write             is too simplistic. In the revised manuscript, I rewrote the discussion and               conclusion, which were added performance assessment and disadvantage           discussion.

12. Comment: In my humble opinion, the “Results” section is mixing the presentation and explanation/description of results, with the description of some techniques used to obtain those results.

         For example, the description of the “Experimental Scheme” and the                      description of the “FSM”, in my opinion, are not results, but part of the                “Materials and methods” used in the experiment.

        So, I suggest moving them into the “Materials and methods” section.

        Response: Your suggestion is of great help to my manuscript. In the                    revised manuscript, I transfer this part to “Materials and methods” section.

Combined with the views of the other two reviewers. I adjusted the clarity          of the picture and added a mRMR algorithm flowchart.

Special thanks to you for your good comments. We are extremely grateful to Reviewer for pointing out this problem. Thank you and all the reviewers for the kind advice again. If you have any questions, please contact us without hesitate.

Yours sincerely

Zeyu Yin

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors addressed my coomnets and improve the manuscript.

Reviewer 3 Report

Dear authors,

Thanks for the corrections.

The paper has been improved a lot.

Regards,