Thrust Vector Observation for Force Feedback-Controlled UAVs

Round 1

Reviewer 1 Report

The paper “Thrust Vector Observation for Force Feedback Controlled UAVs” proposes a scheme for real-time measurement of thrust by load cell placed on the boom. The paper is well-written and clear, it will be interesting for specialists in the UAV control. The proposed engineering solution is worth for consideration for rotors with changing thrust directions. A few minor comments and question should be addressed before the acceptance for publication.

In line 45 the authors state that typically the force model is F=kw^2 but it does not take into account the incoming airflow. It would be interesting to compare the measured value of the force with the computed with this simple model.

The concept has been developed for measuring the two components of the thrust vector in plane perpendicular to the rod. However, the thrust vector can have nonzero component along the rod direction, probably due to some slight damage of the propeller. Is it possible to use the load cell to measure this component as well?

In line 111 the rotation speed dimensions of 1/min is not clear. Is it rmp or rad/min?

No proper estimation of the measurement accuracy is provided in the text. At least, the authors should indicate the value of measurement residuals after the calibration. From Fig. 8 even at 0 m/s inflow and 0 angle it is difficult to estimate the measurement noise – is it about 1N value? Or the observed deviations can be explained by some turbulence effects?

It is not clear how the thrust vector is calculated in right column in Fig.8? Is it just arctan(F_x/F_y)?

Why did the authors rotate the thrust vector only in the range of 0-90 deg? Will the whole setup work for any angle 0-360 deg?

It seems that the effect described in 5.3.1 when the incoming airflow is not aligned with principal axes is the result of beam form-factor. Probably the authors should consider more complex measurement model than just two independent measurement channels.

What caused the angle offset described in 5.3.2? Is it a constant offset or it depends on angle relative to the incoming airflow?

Author Response

# Reviewer1 (plain text)
In line 45 the authors state that typically the force model is F=kw^2 but it does not take into account the incoming airflow. It would be interesting to compare the measured value of the force with the computed with this simple model.
In this research we do not present an alternative approach to the conventional rotor thrust modeling approach. For our goal, to identify a possible solution for coping with inflow induced thrust vector divergence, the modeling of steady state rotor thrust was not necessary. Thus, no rotation rate measurement was performed during the experiments. We consider a comparison with the traditional approach for our future research.
The concept has been developed for measuring the two components of the thrust vector in plane perpendicular to the rod. However, the thrust vector can have nonzero component along the rod direction, probably due to some slight damage of the propeller. Is it possible to use the load cell to measure this component as well?
It might be possible to measure these components, but the noise will probably be higher than the measurement itself.
In line 111 the rotation speed dimensions of 1/min is not clear. Is it rmp or rad/min?
We added the speed dimension (RPM) to the paper.
No proper estimation of the measurement accuracy is provided in the text. At least, the authors should indicate the value of measurement residuals after the calibration. From Fig. 8 even at 0 m/s inflow and 0 angle it is difficult to estimate the measurement noise – is it about 1N value? Or the observed deviations can be explained by some turbulence effects?
The measurement noise can be neglected for the purpose of our measurements, since it is significantly smaller than the noise produced by the inflowing air and the turning servo. A noise level of < 0.5 N was observed in steady tests.
It is not clear how the thrust vector is calculated in right column in Fig.8? Is it just arctan(F_x/F_y)?
Yes, it is calculated by trigonometry using arctan(F_x/F_y).
Why did the authors rotate the thrust vector only in the range of 0-90 deg? Will the whole setup work for any angle 0-360 deg?
0-90 deg is the typical range of motion seen in tilt rotor UAVs. Since we wanted to address the effects which come with tilting a rotor into the direction of travel, only angles up to 90 deg were considered. It is expected, that at angles of attack greater than 90 deg, the thrust deflection will be the same as on the at 90 deg mirrored angle. (but in the downwards direction)
It seems that the effect described in 5.3.1 when the incoming airflow is not aligned with principal axes is the result of beam form-factor. Probably the authors should consider more complex measurement model than just two independent measurement channels.
This would go beyond the scope of this article, since we would need extensive cfd simulations of the propeller, rod and environment of the wind tunnel. Since the airflow at the end of the wind tunnel is not known and turbulent, this will be almost impossible. Future work may include simulations in a more controlled environment.
What caused the angle offset described in 5.3.2? Is it a constant offset or it depends on angle relative to the incoming airflow?
The angle offset is discussed in the referenced chapter 5.3.2 and we provide two possible causes. It is hard to know the exact reason because of the described turbulent air flow.

 

Reviewer 2 Report

Please see the attached file.

Comments for author File: Comments.pdf

Author Response

# Reviewer2 (PDF)   Thank you for the very helpful review! We implemented most of your remarks and really appreciated your feedback.   Don't think you need to capitalize each word of the figure titles Figure 4 typo assambled --> Assembled
Corrected accordingly.
More labeling on each important component in this figure (like the torque rod or inner outer shell) would help readers understand the description and figure better. What is vertical and horizontal? Rotor axis?
Labeling was added to Figure 5. In Figure 7 we added the coordinate axes.
Left, right or center plot?
A remark to the plot was added (right column)
How do you get the inflow speed? Are there any inflow angle measurement or assessment of inflow properties prior to the main test since you didn't use a conventional test section. It is interesting to know the jet size on this outlet and its uniformity and how it would interact with the motor beam.
We clarified how we get the measurement of the inflow speed. The uniform nature of the inflow is discussed within the text. We also added outlet diameter information.
I couldn't find the information on the prop size or the test rpm? Is it fixed rpm through out the whole test or you are aiming for constant force production? With different inflow speed, the prop would start produce different output even with the same rpm.
We added the propeller dimensions in the chapter implementation. The exact test rpm unfortunately could not be measured with the available equipment. We added a sentence regarding the thrust setting and supply voltage. Since no control loop is involved, we did not aim for constant force production. We simply set the motor controller to 60.
So the step increment of the servo angle is in the red line of the right plots? Why is there the big step from 0 to 20 deg?
This was a software bug on testing day. Since it does not affect our conclusion, we decided to keep the data.
You need a diagram showing the force direction with respect to the rotor setup, with positive direction indicators.
In Figure 7 we added the coordinate axes.
I wonder how much of these came from blockage instead of the actual aerodynamic interaction between rotor and inflow?
This is an interesting question, but out of scope for this paper. We will consider this in future work.
How do you quantify noise? 
We referred to the visibly increasing high frequency deviations from the ideal thrust curve.
For blockage effect, you can try to make a measurement with inflow in fixed rotor. The blockage of fixed rotor and spinning rotor wouldn't be the same but at least giving you the ballpark of the rig blockage effect. Uniformity of the inflow also play a big part on the rotor of this size. Seeing the size of the outlet compare to the rotor, I suggest you have a well document information of the inflow when continuing the work further. 
Thank you for the remark! As indicated in our manuscript, the inflow situation was our major problem during evaluation. We will take this into consideration for our future work and conduct our tests at a different facility.
You should put in the summary of your results and its analysis in the conclusion too before proceed to the next step.
The results of the test runs are discussed in the detail in chapter 5.3. Therefore only a short conclusion was given in chapter 6.

 

Reviewer 3 Report

Below is my comment.

Comments for author File: Comments.pdf

Author Response

## Reviewer 3 (PDF)
This paper present Thrust Vector Observation for Force Feedback Controlled UAVs. Below is my comment: 1. What are the advantage of proposed method?
The advantages of the proposed method are discussed in detail in chapter 1,2 and 3. In conclusion the advantages of measuring the true forces on the motor beams are that these measurements can be used in future work by incorporating them in a force feedback loop controller to solve the problem of an externally influenced thrust vector.
2. What is main contribution?
Our main contributions are discussed in detail in the paper - in conclusion: We developed a measurement setup for in-flight thrust vector observation and did a thorough evaluation of the proposed setup at different inflow speeds. We also confirmed the presence of the fundamental problem and validity of our approach. The contributions are listed as bullet points on page 1.

3. The literature review should mention recent method: Active Fault Tolerant Control of a Quadcopter Against Time-Varying Actuator Faults and Saturations using Sliding Mode Backstepping Approach. Applied Sciences
The recommended paper targets the fault tolerant control of traditional quadcopters. In our opinion, citing this work does not improve our paper as required by the MDPI reviewer guidelines:
"Reviewers must not recommend excessive citation of their work (self-citations), another author’s work (honorary citations) or articles from the journal where the manuscript was submitted as a means of increasing the citations of the reviewer/authors/journal. You can provide references as needed, but they must clearly improve the quality of the manuscript under review."
4. The structure of paper should be reorganized. The current form look like a report form.
Our paper follows the proposed structure recommended by the MDPI style guide for a report:
"These are original research manuscripts. The work should report scientifically sound experiments and provide a substantial amount of new information. The article should include the most recent and relevant references in the field. The structure should include an Abstract, Keywords, Introduction, Materials and Methods, Results, Discussion, and Conclusions (optional) sections, with a suggested minimum word count of 4000 words. Please refer to the journal webpages for specific instructions and templates."

 

Round 2

Reviewer 3 Report

This paper is accepted in current form