Back-Support Exoskeleton Control Strategy for Pulling Activities: Design and Preliminary Evaluation

Round 1

Reviewer 1 Report

The authors discuss the use of an active low back exo to assist pulling activities in different conditions. I think that the topic is very interesting and it is valuable in the corresponding scientific field. I do however think that there are some clarification needed before to be considered for publication.

Here some suggestions.

• Please introduce also other exos available in literature and discuss whether and why they may be suitable/not suitable to assist pulling activities.
• The use of the Myo needs to be clarified. The Myo has 8 different bipolar EMG channels – which one(s) do you consider? How do you combine signals of different channels given that they record the activity of different muscles? Given that the MYO measures both flexor and extensor muscles – which movement did you selected to define EMG max?
• Please detail how did you provided the torque to the motor. Was it continuously active when the button was pressed? Which kind of control did you provided to the subject? Torque control at motor side? Did you measured the effective torque at the joint?
• Line 224 - “Evaluating these two metrics allows gaining a complete insight of the changes in muscle activation induced by the exoskeleton assistance.” I do not agree with this sentence, since side effects on other muscles might be observed. Please comment and include in the discussion section.
• 228 - Please include data normality analysis for using ANOVA or use the Friedman test.
• 273 – Please define gammas in the text and not only in the tables’ captions.
• Please include a statistical analysis for rho measurements being statistically below 1, and comment on statistical evidence.
• 288-292. Please provide data variability.

Author Response

Dear Editors and Reviewers:
Thank you for your comments concerning our manuscript. 
Those comments are all valuable and very helpful for revising and improving our work. 
We have studied comments carefully and have made corrections which we hope meet your approval. 
Changes are marked in red on the revised manuscript. 
The main corrections and the responses to the reviewer' comments are as flowing:

- We have introduced some state-of-the-art passive exoskeletons and explained why they are not suitable to assist pulling activities (1. Introduction).
Also, in the Discussion section (4.2. Practical implications), we have given some examples of active back-support exoskeletons and indicated that the strategy proposed in this work can be implemented on other active devices.  

- To control purpose, the sum of the eight EMG signals acquired by the Myo armband was considered.
In fact, we are interested in the overall activity of the forearm muscles, as we need only an estimation and not a precise measure of the grip strength.
The grip strength then is used to define the assistive torques as it is connected to the mass of the object being held during the execution of a MMH task (e.g., lifting, carrying or pulling).
As opposed to acquiring the activity of a specific group of muscles at the forearm, acquiring their overall activity represents a big practical advantage as it eliminates the need for precise electrode placement and calibration. 
Besides that, there are other reasons that make the device particularly convenient for actual use in the workplace: it is affordable, not invasive and it uses dry electrodes, which require no skin preparation nor pre-gelled disposable electrodes.
Finally, the signal is normalized by a maximum value, which is acquired before task execution and can be readjusted quickly during use, if necessary. 
The "calibration phase" to acquire the maximum value is fast and consists of asking the participants to squeeze hard his forearm (sub-maximum exertion task).
Accordingly, we have added in the text more details about this (section 2.1. Control strategy). 
The advantages of the Myo use even for real workplace environment were also summarized in the Discussion section (4.2. Practical implications).

- The assistance is triggered once the button is pressed and the assistive torques were provided continuously for all the time period in which the button was kept pressed. The modulation of the torques followed Equation 1. So for this preliminary stage of the study, the "start" of the assistance is controlled by the experimenter. The subject, on the other hand, has control over the modulation of the assistive torques by contracting his forearm muscles.
For future works and real scenario applications, we will find other solutions (e.g., place the button into the tool the user uses for pulling or into ad-hoc gloves) because the user should have direct access to the trigger command. 
The control scheme is structured in three levels. So the torque as defined by Equation 1 is the output of the middle level and it is the reference signal that the low level aims to track. The output torques of the two current-controlled actuators are regulated by the low level, tracking the reference signal defined by the middle level with a close-loop proportional-derivative (PD) controller. For each side, the feedback torque is measured with a strain gauge-based torque sensor.
Detailed descriptions of the control system are available in the references [14,33].
As regards the torque values, the ones indicated in the manuscript are the effective torques at the joint, as measured by the torque sensors. 
The low-level controller of the exoskeleton and how it tracks the torque reference defined by the middle level is outside the scope of this work and is therefore only briefly described.
Figure 1 shows for one participant the torque reference signal and the torque provided to the user, as measured by the torque sensors.
The torque tracking capability of the exoskeleton low-level control depends on the actuator physical characteristics, as its maximum speed and its dynamic performance. The assistive torque follows the reference signal with a certain bandwidth (which depends on the reflected motor inertia) characterized by a single dominant pole; thus, at increasing velocity the torque tracking capability decays. 
With respect to the previous version of the exoskeleton (see [16]), there was an improvement in the tracking accuracy and the applied assistance (torque output) properly follows the reference signal (torque ref). 
On the other hand, the errors in the reference torque (Figure 1) do not have an influence on the applied torque.

Figure 1

- Line 224 - (line 236) “Evaluating these two metrics allows gaining a complete insight of the changes in muscle activation induced by the exoskeleton assistance.” I do not agree with this sentence, since side effects on other muscles might be observed. Please comment and include in the discussion section.
This is true, we forgot to add SPINAL muscle activation.
Change accordingly: "Evaluating these two metrics allows gaining a complete insight of the changes in spinal muscles activation induced by the exoskeleton assistance."
We have also added the concern about side effects on other muscles in the discussion section.

- Change accordingly: (line 240) "The RMS and the 90th percentile muscle activities (i.e., the dependent variables) are statistically tested using three-way repeated measures ANOVA to study the effects of the multiple factors (i.e., the independent variables): assistance mode (no-exo and exo), pulling height (waist and shoulder), box weight (10 and 20 kg), and their interactions. 
To perform ANOVA analysis, the normality of the distributions of the dependent variables was tested with the Kolmogorov-Smirnov test (at the 5% significance level)".

- Change accordingly: (line 286) "In Tables1 and 2 the numbers of γ- and γ+ (i.e., number of subjects γ- for which the exoskeleton use results in reduced muscle activity (ρi < 1) and number of subjects γ+ for which the exoskeleton use results in increased muscle activity (ρi > 1)) are reported for the two metrics separately and for each combination of pulling height and box weight".

- We performed statistical analysis to find statistical differences between rho and distribution of 1. These differences were statistically significant for RMS and 90th percentile EMG values in condition waist height 10 kg and only for RMS EMG value in condition shoulder height 10 kg.
Results were added in the text. 
As already discussed in the related section (4.1. Effects of the assistance on muscle activity and subjective measurements) the results were different between the two different payloads, and specifically greater reductions occurred for tasks executed with the 10 kg box for which we have also found statistical significance. This is probably due to the magnitude of the provided torque, i.e. for the heavier load the torque provided was not enough to obtained significant reductions of the activity of the back muscles.  

- Data variability added (lines 305-309). 

 

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper is an useful paper for exoskeleton researchers. In particular, this paper presents well the evaluation of the effectiveness.

 

However, this paper needs to explain about the reliability of sEMG. Of course, this experiment is executed in the laboratory but, versatility is one of the most important issues. 

Author Response

Dear Editors and Reviewers:
Thank you for your comments concerning our manuscript. 
Those comments are all valuable and very helpful for revising and improving our work. 
We have studied comments carefully and have made corrections which we hope meet your approval. 
Changes are marked in red in the revised manuscript. 

- Thank you again for your feedback.
Accordingly, we have added in the text more details about the myo device and how it works (section 2.1. Control strategy) and its practical advantages and convenience for being used in a real industrial environment (section 4.2. Practical implications). 

The Myo armband integrates eight pairs of dry electrodes that record via surface electromyography (sEMG) the activity of the forearm muscles.
To control purpose, we considered the sum of the eight EMG signals acquired by the Myo armband, as we are interested in the overall activity of the forearm muscles as an estimation of the grip strength.
The grip strength then is used to define the assistive torques as it is connected to the mass of the object being held during the execution of the task.

There are many reasons that make this device particularly convenient for actual use in the workplace: it is affordable and easy-to-use, is powered by built-in batteries, uses wireless communication to send out data, is not invasive or uncomfortable to wear, and uses dry electrodes, which require no skin preparation nor pre-gelled disposable electrodes.
Moreover, as opposed to acquiring the activity of a specific group of muscles at the forearm, acquiring their overall activity represents a big practical advantage as it eliminates the need for precise electrode placement and calibration. 

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper is very well structured and detailed from formal analysis to experimental tests.

The paper present an extensive introduction for assiste trunk exoskeleton but something more should be detailed for trunk characterization using wearable sensors that can help also while the exoskeleton is in use, this reference can help to strength this part of the paper:

 Please better detail how  Subjective measurements is done

Cafolla, D., Chen, I.-M., Ceccarelli, M. An experimental characterization of human torso motion (2015) Frontiers of Mechanical Engineering, 10 (4), pp. 311-325. https://www.scopus.com/inward/record.uri?eid=2-s2.0-84949535136&doi=10.1007%2fs11465-015-0352-z&partnerID=40&md5=b556581a90163afff5f75b278446a795 DOI: 10.1007/s11465-015-0352-z

After this minor revision this paper is valuable of publication in my opinion.

Author Response

Dear Editors and Reviewers:
Thank you for your comments concerning our manuscript. 
Those comments are all valuable and very helpful for revising and improving our work. 
We have studied comments carefully and have made corrections which we hope meet your approval. 
Changes are marked in red in the revised manuscript. 

- According to your suggestion, we have added in the text the need of characterize the trunk motion while the exoskeleton is in use, as it can be useful also for adjusting the assistive torques accordingly (4.3. Limitations): "Additional assistance may also be required to support users because of the adopted posture. For instance, the trunk motion during the execution of pulling tasks while wearing the exoskeleton can be further characterized to modulate the assistance accordingly [62])."

Author Response File: Author Response.pdf