Design, Construction and Validation of a Proof of Concept Flexible–Rigid Mechanism Emulating Human Leg Behavior

Round 1

Reviewer 1 Report

The study is interesting and well performed. I suggest some improvements to be considered for the revision. Additional session regarding results and discussions can be added. The current one was too short to adequately understand the performance of the modeling, design and control. The crucial parameters are not clear for the proposed flexible-rigid manipulator in emulating the lower limb kinematics? Limitations or assumptions for the current design are lacking or inadequate. Results should be discussed with similar researches in order to understand the contribution of the study.

Author Response

Response to the reviewers:

 

We would like to thank the reviewers for their constructive comments. The corrections suggested to us have been implemented and are detailed below.

 

Response to Reviewer 1 Comments

Point 1: The study is interesting and well performed. I suggest some improvements to be considered for the revision. Additional session regarding results and discussions can be added. The current one was too short to adequately understand the performance of the modeling, design and control. The crucial parameters are not clear for the proposed flexible-rigid manipulator in emulating the lower limb kinematics? Limitations or assumptions for the current design are lacking or inadequate. Results should be discussed with similar researches in order to understand the contribution of the study.

Response 1: We would like to thank the reviewer for the suggestions. We have considered your comments and added several new sections:

 

• We added a new section called “Results & Discussions” P9 L 258-293 that interprets the results of the proposed design pipeline.

 

• We added a paragraph discussing the limitations of the proposed design pipeline on P5 L125-130.

 

• We added “Table 1” on P6 L179 which includes all of the muscle specific parameters used for simulation. The desired kinematics were obtained by each participant and muscle activations were found by the simulation that guided the excitation or actuation pattern. We explained it better at the beginning of section 4.1 P7 L196-200.

 

• This was an excellent point and we have added an additional set of complementary plots (Figure 8bd) and expanded our validations into the Results & Discussion section to address the lack of quantitative analysis to compare Human vs. Prototype & Human vs. OpenSim 3.0 Model. 

 

• The Results and discussions section explains how the current approach is different from the one chosen by other researchers.

Author Response File: Author Response.pdf

Reviewer 2 Report

Title
The term “proof of concept” should be reported in the title.

Introduction

P2 L38: The difference between the present work and reference [26] (same team) is not clear. Introduction should be completed to demonstrate the novelty of the present work.

P2 L52-60: The need for a new tensegrity simulation environment is not clear as the limits of existing solutions are not reported.

P2 L61-62: The aim of the paper is not clear: development and assessment of a design methodology (suggested in the introduction), or development and assessment of a model (or a tensegrity manipulator)?

Simulation Modeling

It is not clear if the Opensim model was designed based on an existing tensegrity manipulator or if the Opensim model helps to design the tensegrity manipulator.

P3 L71: Which version of Opensim was used? This information should be reported here.

P3 L71-72: It seems that the Opensim model was only used to define the manipulator control. Thus, why section 2 has been reported before section 3? Section 3 would allow to define the manipulator before to establish a methodology to control it.

P3 L72-73: How were defined the model geometry, the muscles path …? How did you ensure that real and model geometries were similar?

P3 L74-75: It is not clear why different models have been generated? It is to match the dimension of a participant on who gait patterns have been recorded?

P3 L75-86: It is not clear why the forward dynamics tool was used to generate muscle activation patterns. Indeed, based on recorded desired kinematics, the common tool to compute the muscle activations that allowed the recorded motions is the static optimisation tool, i.e. a tool based on inverse dynamics. At the opposite, FD tool allows to compute the motions resulting from a set of muscle activations … Please revise this section to better explain what has been done to compute the muscle activations related to a desired gait pattern.

P4 Eq3: What means the “.” at the beginning of the equation?

P4 L81-86: How were set the muscle parameters maximal isometric force, optimal fiber length, tendon slack length, pennation angle? Indeed, these parameters are crucial for a correct use of the model. These values should be reported in the paper.

P4 L87-89: How active tensile element lever arms have been adjusted in the model to fit the tensegrity manipulator ones? Indeed, if the model lever arms are different than in the manipulator, the resulting muscle forces will also be different.

Structural design

The tensegrity manipulator design has been developed in an older study [26]. This should be clearly reported here.

P6 L98-99: It is not clear why the compression element lengths have been adjusted to the Opensim model ones while the Opensim model is then scaled to various dimensions.

P6 L130-131: Why the authors used in-place gait motion rather than normal gait? This movement is rarely explored and the muscles activation seems quite different than during normal gait.

Sections 3.4 and 3.5: How the actuators were sized, and/or how the cable paths have been defined to ensure sufficient actuator force to generate a desired motion? Did you used Opensim to assess required forces at this stage?

Control and Actuation Strategies

P7 L162: If I understand well, desired kinematics was obtained on a participant, and muscle activations using the static optimisation tool. The resulting activations were used as initial guess in the FD tool to manipulate hip and knee joints. If so, Figure 4 is not clear enough as well as section 4.1.

Figure 5: “Applied contracting forces” should be replaced by “applied activations”.

P8 L178: What is “section A”?

P8 L204-207: This part should be moved in Section 3 to explain the motor sizing.

Gait Experiment

Ethical authorisations are not reported here.

P10: Static optimisation tool instead of FD tool (see previous comments).

Figures 8 and 9: Some titles are partly hidden.

Conclusion

The validation proposes here is only qualitative. It would be much more relevant with well chosen metrics that would demonstrate quantitatively the ability of the model/manipulator to reproduce recorded motions.

Results are not discussed nor compared with previous literature.

Limitations of the study are not reported not discussed.

Author Response

Response to the reviewers:

 

We would like to thank the reviewers for their constructive comments. The corrections suggested to us have been implemented and are detailed below.

 

Response to Reviewer 2 Comments

Point 1: The term “proof of concept” should be reported in the title.

 

Response 1: Thank you for your suggestion, we have changed the title of the paper to include proof of concept. The new title is: “Design, Construction and Validation of a Proof of Concept Flexible-rigid Mechanism Emulating Human Leg Behavior”.

 

Point 2: P2 L38: The difference between the present work and reference [26] (same team) is not clear. Introduction should be completed to demonstrate the novelty of the present work.

 

Response 2: We would like to thank the reviewer for noticing that we have not specified the differences between the older conference paper and the new journal submission. We have added more detail starting on P3 L67. In the revised version of the paper we are explaining that the current journal paper uses the conference paper as a starting point and proposes an improved mechanical design and sensing strategy (e.g., IMU localization for monitoring the manipulator's dynamic behavior). One of the main contributions of the journal paper is the pipeline that combines the physical prototype and human subject in the same environment. This approach could be used for streamlining the development of future flexible mechanisms.

Point 3 P2 L52-60: The need for a new tensegrity simulation environment is not clear as the limits of existing solutions are not reported.

 

Response 3: We have included additional details added to P2 L55-66 to address the current limitations of existing solutions and how OpenSim is capable of bridging the gap between robotic and human models.

 

Point 4: P2 L61-62: The aim of the paper is not clear: development and assessment of a design methodology (suggested in the introduction), or development and assessment of a model (or a tensegrity manipulator)?

 

Response 4: The focus of the paper is proposing a pipeline for developing a flexible-rigid robotic system that replicates the musculoskeletal connections within the human leg and can replicate aspects of the human leg kinematic and dynamic behavior. We have reworded P3 L70-73.

 

Point 5: It is not clear if the Opensim model was designed based on an existing tensegrity manipulator or if the Opensim model helps to design the tensegrity manipulator. P3 L71: Which version of Opensim was used? This information should be reported here.

 

Response 5: We developed the robotic manipulator and the OpenSim model at the same time. The goal was for the model to inform the structural design and therefore, predict its dynamic behavior. We clarify our approach on P3 L79.

 

We used OpenSim 3.0. We agree with the reviewer that it is important to report the version of OpenSim, and therefore, referred to the package as “OpenSim 3.0” throughout the paper.

 

 

Point 6: P3 L71-72: It seems that the Opensim model was only used to define the manipulator control. Thus, why section 2 has been reported before section 3? Section 3 would allow to define the manipulator before to establish a methodology to control it.

 

Response 6: Thank you for the suggestion, we reorganized the section by introducing the design (Section 2) before the simulation (Section 3).

Point 7: P3 L72-73: How were defined the model geometry, the muscles path …? How did you ensure that real and model geometries were similar?

 

Response 7: Thank you for bringing this to our attention. We have addressed the concern in P5 L164-169. We explained that the real and model geometries are proportionally matched using the Scaling tool and reflective infrared markers. 

 

Point 8: P3 L74-75: It is not clear why different models have been generated? It is to match the dimension of a participant on who gait patterns have been recorded?

 

Response 8: We agree that the reviewer has a valid concern where different models were generated to demonstrate the calibration process only required a static file with markers that match the same placement on the simulated model and the physical subject. We have added more detail to P5 L158-168. 

 

The answer to the second question is correct and we use this to adapt to differently sized models.

 

Point 9: P3 L75-86: It is not clear why the forward dynamics tool was used to generate muscle activation patterns. Indeed, based on recorded desired kinematics, the common tool to compute the muscle activations that allowed the recorded motions is the static optimisation tool, i.e. a tool based on inverse dynamics. At the opposite, FD tool allows to compute the motions resulting from a set of muscle activations … Please revise this section to better explain what has been done to compute the muscle activations related to a desired gait pattern.

 

Response 9: We agree with the reviewer that a similar result could be obtained using the Static Optimization tool in OpenSim. However, in our approach we used the Forward Dynamics tool. One of our goals was to demonstrate that for a set of desired kinematics, we can iterate through combinations of muscle activations to yield the required forces/activation patterns. This is now clarified in the Simulation Modeling section starting on P6 L173

 

Point 10: P4 Eq3: What means the “.” at the beginning of the equation?

 

Response 10: Thank you for finding that issue, that is a typo and has been removed from P6 Eq3.

 

Point 11: P4 L81-86: How were set the muscle parameters maximal isometric force, optimal fiber length, tendon slack length, pennation angle? Indeed, these parameters are crucial for a correct use of the model. These values should be reported in the paper.

 

Response 11: Thank you for your suggestion, we have added Table 1 on the top of P7 for the muscle element specific parameters used for simulation.

 

Point 12: P4 L87-89: How active tensile element lever arms have been adjusted in the model to fit the tensegrity manipulator ones? Indeed, if the model lever arms are different than in the manipulator, the resulting muscle forces will also be different.

 

Response 12: Each of the tensile elements match the exact placement of the robotic manipulator using the Scaling tool to proportionally match compression elements and tensile connections. We have added more information to P6-7 L186-190 to clarify. 

Point 13: The tensegrity manipulator design has been developed in an older study [26]. This should be clearly reported here. P6 L98-99: It is not clear why the compression element lengths have been adjusted to the Opensim model ones while the Opensim model is then scaled to various dimensions.

Response 13: Thank you for bringing this up to our attention, this should have been addressed by P2 L67-70, and P5 L163-173 where each OpenSim model proportionally matches the physical properties using the Scaling tool.

 

Point 14: P6 L130-131: Why the authors used in-place gait motion rather than normal gait? This movement is rarely explored and the muscles activation seems quite different than during normal gait.

Response 14: Thank you for asking this crucial question, we recognize that normal gait requires a complex design and due to this being a proof of concept tensegrity-inspired design we wanted to focus our attention on aligning the biomechanics and robotic manipulators within the same system and prove that it is capable of more. Please see P5 L125-130.

 

Point 15: Sections 3.4 and 3.5: How the actuators were sized, and/or how the cable paths have been defined to ensure sufficient actuator force to generate a desired motion? Did you used Opensim to assess required forces at this stage?

 

Response 15: We used OpenSim to assess the required forces and activations as well as help with actuator design and selection. Please see where we added more details in the Introduction Section P3 L79-83, in the Simulation Modeling section and additionally in P5 L152-154.

Point 16: P7 L162: If I understand well, desired kinematics was obtained on a participant, and muscle activations using the static optimisation tool. The resulting activations were used as initial guess in the FD tool to manipulate hip and knee joints. If so, Figure 4 is not clear enough as well as section 4.1.

 

Response 16: This is exactly our intention that the desired kinematics were obtained by each participant and muscle activations were found by the simulation that guided the excitation or actuation pattern. We explained it better at the beginning of section 4.1 P7 L196-200.

 

Point 17: Figure 5: “Applied contracting forces” should be replaced by “applied activations”.

 

Response 17: Thank you for bringing this up to our attention, we have made the proper changes to Figure 5.

Point 18: P8 L178: What is “section A”?

 

Response 18: Thank you for finding this issue, we have changed it to section 4.1.

 

Point 19: P8 L204-207: This part should be moved in Section 3 to explain the motor sizing.

 

Response 19: Thank you for bringing this suggestion to our attention, we have moved the part accordingly P5 L158-161.

 

Point 20: Ethical authorisations are not reported here.

 

Response 20: We have added a section including the ethical autorizations regarding the IRB Exemption.

 

Point 21: P10: Static optimisation tool instead of FD tool (see previous comments).

 

Response 21: Thank you for bringing this to our attention, we agree that the static optimization tool would be beneficial in most applications.

 

Point 22: Figures 8 and 9: Some titles are partly hidden.

 

Response 22: Thank you for bringing this to our attention, we have reformatted the Latex to fix this issue for Figure 8 & 9.

 

Point 23: The validation proposes here is only qualitative. It would be much more relevant with well chosen metrics that would demonstrate quantitatively the ability of the model/manipulator to reproduce recorded motions.

 

Response 23: This was an excellent point and we have added an additional set of complementary plots (Figure 8bd) and expanded our validations into the Results & Discussion section to address the lack of quantitative analysis to compare Human vs. Prototype & Human vs. OpenSim 3.0 Model.

 

Point 24: Results are not discussed nor compared with previous literature.

 

Response 24: Thank you for bringing up this point, we have added a new section “5. Results and Discussions” starting at P9 L 258.

Point 25: Limitations of the study are not reported not discussed.

Response 25: Thank you for the advice. We discussed the limitations of our system on P5 L125-130 and we added a new section “5. Results and Discussions”. P9 L 258-293

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper can be accepted in the current version.

Reviewer 2 Report

The authors have provided clear responses to each comment and have sufficiently addressed the major concerns raised. The revised manuscript is significantly clearer.