Robust Data Fusion of UAV Navigation Measurements with Application to the Landing System
Round 1
Reviewer 1 Report
The authors have reviewed the work with great superficiality: the feeling is that they have worked quickly: there are typing errors, spelling errors and often the English style is poor and difficult to understand. Many sentences are long and convoluted, and punctuation isn’t often used correctly. This leads to a difficulty in reading some parts of the document.
Below some findings:
line 23: sensisng must be corrected in sensing
line 24: othen must be corrected in often
line28: hard applicable must be corrected
line 29: continuous objects?
line 33-35: methods is repeated five times
line 39: to identifying must be corrected
line 49: superior is not the right adjective
line 56: airfiled must be corrected to airfield
line 90: works should be changed
line 114: It's the worst sentence to highlight the paper originality..and is also misspelled: ..contribution fo our work
line 127: "is" lacks?
line 138: "for" lacks?
line 145: "On" must be corrected to "in"
line 184: "given" is incorrect
The narration is often confused and also the reference to previous works is too frequent and the mode in square brackets too long. The proposed method shows ideas of originality, but the paper needs a good review.
Author Response
The authors have reviewed the work with great superficiality: the feeling is that they have worked quickly: there are typing errors, spelling errors and often the English style is poor and difficult to understand. Many sentences are long and convoluted, and punctuation isn’t often used correctly. This leads to a difficulty in reading some parts of the document.
Below some findings:
line 23: sensisng must be corrected in sensing
Fixed
line 24: othen must be corrected in often
Fixed
line28: hard applicable must be corrected
The sentence corrected
line 29: continuous objects?
More explanation is given in text
line 33-35: methods is repeated five times
Fixed
Thanks for careful reading, we tried to correct and amend all places remarked by reviewer.
Reviewer 2 Report
The paper addresses the difficulties commonly found when performing automatic landing of UAV: how to fuse the measurement coming from different sensors and obtain the best estimation for the controlled variables. Manuscript’s proposal is based on a novel algorithm under the least modulus method over the Kalman filter, augmented with a chi-square test to detect prediction error and calculate a better estimation.
MAJOR ISSUES:
- The main concern comes from the fact that the proposed method is only tested under simulations, but not validated in real-world-scenarios (or laboratory).
- The research seems to be not complete, or unfinished. Certain well-known solutions found in the literature are used, but finally I cannot see how it faces all issues that arise.
- The dynamic of the flight is not addressed, and it should be taken into consideration.
- The title projects and idea of what the paper addressed (automatic landing of an UAV), but the manuscript just focuses on a part of that problem.
- Extensive English edition is required. Some sentences are at least fuzzy and difficult to read.
- The state of the art:
- Some sentences need support in the literature (as “The only advantage of machine learning methods“ in line 36, and others).
- It should be updated with more recent manuscripts. It is acceptable to use reference from the 80’s and 90’s, but this issue has been researched in the present as well, and the reader would appreciate understanding how others face the same problem.
- Could have references related to the environment of the problem (automatic landing).
- The problem statement description should be elaborated some more, and the relationship with the title of the manuscript should be stated.
- The methods should be described deeply: how are the simulations performed?
MINOR CONCERNS:
- Some minor spelling issues should also be addressed (like ‘sensisng’ in line 23, ‘getiing’ in 24, and many others).
Author Response
MAJOR ISSUES:
- The main concern comes from the fact that the proposed method is only tested under simulations, but not validated in real-world-scenarios (or laboratory).
- This conclusion is probably result in rather short description of our testing results which made with the real flying UAV and real system of sensors either on-board or terrain based.
The additional paragraphs in Section 5 expand the description.
- The research seems to be not complete, or unfinished. Certain well-known solutions found in the literature are used, but finally I cannot see how it faces all issues that arise.
- We stress that the aim of research was the development of robust estimation algorithm in the case of abnormal errors. For details one should see the Section 4.
- The dynamic of the flight is not addressed, and it should be taken into consideration.
- Of course, the dynamics of flight is very important and it was taken into account in the System model 4.1 and Model of measurements 4.2 and in the development of the robust estimation algorithms. Meanwhile we should stress that the real flights were made by standard program with standard measurement INS.
- The title projects and idea of what the paper addressed (automatic landing of an UAV), but the manuscript just focuses on a part of that problem.
The task of UAV navigation is quite large, but we only focus on solving the landing part of the flight. Our goal was simply to accompany the landing with robust measurement and estimation algorithms in the presence of abnormal errors in the altitude and locator measurements.
- Extensive English edition is required. Some sentences are at least fuzzy and difficult to read.
Done
- The state of the art:
- Some sentences need support in the literature (as “The only advantage of machine learning methods“ in line 36, and others).
Done
- It should be updated with more recent manuscripts. It is acceptable to use reference from the 80’s and 90’s, but this issue has been researched in the present as well, and the reader would appreciate understanding how others face the same problem.
Robust filtering is rather long-standing problem and the principal references had been done. The problem of abnormal outliers particularly in relation to the UAV position estimation is more recent. The list of references related to the fata fusion for the UAV INS is expanded.
- Could have references related to the environment of the problem (automatic landing).
References related to the automatic landing are presented in the list of References and they are rather specific for large UAV using the runways.
- The problem statement description should be elaborated some more, and the relationship with the title of the manuscript should be stated.
- We suppose that the problem description is elaborated in full details.
- The methods should be described deeply: how are the simulations performed?
- The simulation of stochastic system is a standard process in mathematical modelling. All necessary details related to the problem under consideration are in Sections 3-4
Round 2
Reviewer 2 Report
At this point, my main concerns are focused on:
1.-Answer to reviewer:
a). Some questions rised by reviewer are answered with a simple "Done" (sentences that need support in the state of the art review, extensive English edition);
b). Others simply with vague description ("one should see the Section 4", "All necessary details related to the problem under consideration are in Sections 3-4");
c). Or directly sidelined ("We suppose that the problem description is elaborated in full details")
Please, provide an answer that allows the reader understand what has been changed, why, and where.
2.-English writing-style: please, consider reaching a native english writer that can provide proof-writing. The overall English style should be revised.
Author Response
The article is carefully edited by English speaking editor. So the large part of the previous version is completely rewritten. These principal moments are discussed in the chapter 5. Discussion of test results where we describe our approach to the development of robust data fusion algorithms of the UAV position estimation. The details of the mathematical approach such as the System model, Model of measurements and Data fusion algorithms are presented in Sections 4.1-4.3.
The essence of the approach is the usage together of Kalman Filtering and L1 minimization which permits to localize the abnormal errors and to minimize their influence on the accuracy of UAV position estimation. The switching from one type of estimation to another is based Chi-squared statistical test which is another novelty of the approach.
In order to understand the editorial effort we attached the file with the editor remarks which were carefully revised.
1.-Answer to reviewer:
a). Some questions rised by reviewer are answered with a simple "Done" (sentences that need support in the state of the art review, extensive English edition);
b). Others simply with vague description ("one should see the Section 4", "All necessary details related to the problem under consideration are in Sections 3-4");
c). Or directly sidelined ("We suppose that the problem description is elaborated in full details")
Please, provide an answer that allows the reader understand what has been changed, why, and where.
2.-English writing-style: please, consider reaching a native english writer that can provide proof-writing. The overall English style should be revised.
