Design and Passive Training Control of Elbow Rehabilitation Robot

Round 1

Reviewer 1 Report

Dear Authors,

You have built a prototype for elbow movement in rehabilitation. Pneumatic actuators drive it. Your prototype exists only at your laboratory, but its uniqueness is very doubtful. Especially because you forgot to distinguish it from other prototypes in the manuscript. You present your control, but you do not compare it to another controller. Overall, it remains questionable why you use an ADRC at all. The advantage is not clear in your manuscript. I have some comments for your manuscript that you should address before I can recommend acceptance.

Comments for the revision of the manuscript:

• You introduce with a very general paragraph. However, you do not have a single reference that can confirm your statements in the first paragraph.
• You need to enhance significantly lines 47 to 54. Provide much more detail on the design and control of other prototypes. From this, it must become clear what is special about your design and control. For each reference, write what is important about it.
• After line 96, you do not have a single citation. That needs to be improved. EXAMPLE: You should compare your results with the results of other prototypes in literature.
• For the mechanical design I lack a detailed requirements analysis. Among other points, you should answer which torque you have to provide for the application, which speeds are necessary and what your technical safety concept is to prevent unphysiological movements.
• Figure 1 should be divided into functional groups by color. You should also provide a legend for these groups.
• In Figure 2 I see a lot of whitespace. The presentation is not very clear, additional drawing space could be helpful, use all the space you have.
• In line 132 you use strange commas
• Throughout the manuscript, you must make sure that the spacing between the measure and the unit is set properly.
• From line 141 to 154, I am missing text on the many equations. Write in whole sentences. Better even in several sentences...
• Where do the numerical values in equation 12 originate?
• Equations 15, 16 and 19 are very similar.
• Equation 18 is inadequately addressed in the written text.
• When you write about uncertainty in line 254, I expect you to give a statement, substantiated with references, in which range the load is and varies.
• Equation 20 does not seem correct to me. You did not find the linear extended state observer by yourself. Be sure to indicate your reference source. I recommend that you then also use the notation of the source.
• In table 2 you specify controller parameters. Answer which tuning algorithm you used and give a source if possible. Here I also expect a brief overview of why you have chosen this controller architecture.
• Figure 3 and 4 should scale them so that the timeline goes to maybe 4 or 5 seconds. This way you waste a lot of space. Use the subplot function.
• What software did you use for the simulation, give a precise specification.
• You use commercial purchased parts in your prototype, make a clear identification to all bought-in parts.
• Take care of your units. If you use "°", you should not use "degree".
• The reference in line 336 does not seem correct.
• Where did you get the values in equation 24. Be sure to include a reference.
• A commentary on all result and simulation plots: Export your plots as PDF so that they are inserted as vector graphics. Design the individual figures so that all axis labels and legends are easily readable (perhaps a slightly smaller font than the text part). Avoid Chinese characters in all plots.
• Use $\theta_d$ instead of $theta_d$. This must be consistent.
• Figure 6 a-c has the same caption. Figure 6 b and c look very similar. Revise this figure. Use the subplot function.
• You should explain the movement function in line 407 in more detail. What forms of exercise are commonly used in rehabilitation? Here I expect proper literature work and references from you.
• Why are you doing the 5° step?
• A causal system is more likely to lag behind the reference. How exactly do you estimate the hysteresis in lines 424-425?
• What is the gender of the test subjects in Table 4?
• When you talk about low, medium and high speed, what are the maximum deflections and speeds?
• The most important Last: I expect you to clearly contrast your design against existing prototypes. You also need to compare your control strategy against samples from the literature. You should also see how the PID controller works without an observer. I expect more comparison and discussion here.

I suggest major revisions for your manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper proposes a design and passive training control of a robot driven by artificial muscles for rehabilitation purposes. The topics of the paper are interesting. However, the overall quality of the paper is low, and several points need to be considered to improve the manuscript.

1. The main contributions of the paper with respect to the present literature are not clear. Authors should highlight the novelty and originality of the work.
2. The abstract is not clear. The first sentence of the abstract is 6 rows long. Please highlight the main contributions and results of the work with clear and concise sentences.
3. The results are quite poor. I suggest introducing quantitative metrics (for example root means square error and/or mean absolute error) to better evaluate the performance of the prototype and the tracking error.
4. It is not clear how the PID parameters have been tuned.
5. The quality of the figures is low. I suggest replacing the graphs with high quality images.
6. English should be improved and the whole manuscript should be checked again for typos.
7. Line 115. Elbow rehabilitation robot modeling3. Results. Please check.
8. The following references are suggested to improve the literature review:

Ball, S. J., Brown, I. E., & Scott, S. H. (2007, September). MEDARM: a rehabilitation robot with 5DOF at the shoulder complex. In 2007 IEEE/ASME international conference on Advanced intelligent mechatronics (pp. 1-6). IEEE.

Niu, J., Yang, Q., Chen, G., & Song, R. (2017, July). Nonlinear disturbance observer based sliding mode control of a cable-driven rehabilitation robot. In 2017 International Conference on Rehabilitation Robotics (ICORR) (pp. 664-669). IEEE.

Scalera, L., Gasparetto, A., & Zanotto, D. (2019). Design and experimental validation of a 3-dof underactuated pendulum-like robot. IEEE/ASME Transactions on Mechatronics, 25(1), 217-228.

Liang, G., Ye, W., & Xie, Q. (2012). PID control for the robotic exoskeleton arm: Application to rehabilitation. In Proceedings of the 31st Chinese Control Conference (pp. 4496-4501). IEEE.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript proposes the Design and Passive Training Control of Elbow Rehabilitation Robot with comparative tests on two different control heuristic: ADRC and PID. The ADRC approach leads to a control strategy aimed to remove disturbance to improve performance. The main contribution of this paper is indeed related to reporting the results comparing the diverse performance of the two controllers relatively to the exoskeleton system.

The paper presents an interesting investigation and a relevant topic for this journal. Accordingly, the authors proposed a set of tests to create a discussion regarding the performance report, the evaluation, and the applicability of the proposed system to users. There are some points to be addressed to improve the presented work that would be great to be addressed as a minor revision.

1. The authors propose a comparison between ADRC and PID controllers. Despite such a comparison is very interesting and ends with coherent results, it is recommended that the authors can elaborate on why they chose a PID controller against ADRC. A comparison with other strategies (involving, for example, extended Kalman filtering or model predictive approach) could have been more significant for the analysis.
2. The introduction clarifies the motivations of this work and the social relevance of the presented research. However, it is recommended that the authors can also provide a substantial number of reference reporting researches investigating other control strategies addressing similar problems on rehabilitation exoskeletons.
3. The authors should enrich conclusions with considerations regarding the results of the work.
4. Further proofreading would improve the overall quality of the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear Authors, 

I notice that you have made several revisions. Regarding some comments from round 1, some supplements are needed. I think I expressed myself a somewhat unclearly. If I ask you something in the review report, you are not supposed to answer me directly in the coverletter, but to include the information in your paper. Accordingly, please respond to the following comments: 

• "Throughout the manuscript, you must make sure that the spacing between the measure and the unit is set properly." For example, you still write in line 89-90 measure space unit. In line 126 you have forgotten the space. CHECK YOUR ENTIRE MANUSCRIPT ACCORDINGLY. 
• "Figure 3 and 4 should scale them so that the timeline goes to maybe 4 or 5 seconds. This way you waste a lot of space. Use the subplot function." After second 4, the signals in the plots are very constant. Adjust the time axes accordingly. With the adjusted timelines, the three plots fit easily next to each other horizontally. The resolution of the illustrations is not appropriate for a journal publication. 
• "What software did you use for the simulation, give a precise specification."
• "A commentary on all result and simulation plots: Export your plots as PDF so that they are inserted as vector graphics. Design the individual figures so that all axis labels and legends are easily readable (perhaps a slightly smaller font than the text part). Avoid Chinese characters in all plots." All figures from Matlab should now be vector graphics. The labels should be slightly smaller than the font of the text part.
  -> Revise Figure 3, 4, 5, 6, 7, 8, 10, and 12.
• "What is the gender of the test subjects in Table 4?"
• "When you talk about low, medium and high speed, what are the maximum deflections and speeds?"
• "The most important Last: I expect you to clearly contrast your design against existing prototypes. You also need to compare your control strategy against samples from the literature. You should also see how the PID controller works without an observer. I expect more comparison and discussion here. I suggest major revisions for your manuscript.
  Response 30: Thank you for your conclusive suggestion. We regret not to compare with some other intelligent control methods. For future research, we will consider extended Kalman filtering or model predictive approach applying to elbow joint rehabilitation robots. "
  You are not the first to build an control an elbow rehabilitation system.  Clarify better what makes your work unique.

In total I recommend minor revisions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper was improved as suggested.

Author Response

Thank you again for your comments on the paper. The paper is improved under your suggestions significantly.