A Novel Adaptive Factor-Based H∞ Cubature Kalman Filter for Autonomous Underwater Vehicle
Round 1
Reviewer 1 Report
The introduction gives a clear and broad overview of filtering techniques with the aim of underwater navigation. The paper itself describes an improved adaptive H inifinity Cubature Kalman Filter by adding a fading factor. The method of the A-H-inf-CKF was published by the authors in "Mathematical Problems in Engineering" with almost the same title (A Novel Adaptive H-Infinity Cubature Kalman Filter Algorithm Based on Sage-Husa-Estimator for Unmanned Underwater Vehicle", but this paper shows the achievements by extenting the filter to the AF-H-inf-CKF. With just slightly increasing calculation time the estimation errors can be decreased significantly. The presentation of the theory is precisel and comprehensable without being too detailed. The only drawback I am able to mention is the experiment. You talk about underwater vehicles and validate with a surface test. I understand the problem that underwater tests are almost impossible without huge test facilities, but the dynamics of underwater motion without periodic disturbances from waves or wind is much lower becoming more difficult to predict. I also would like to see measurements of longer periods, e.g. one hour or ten hours, but this can't be done within a couple of weeks, so this is an advice for future work.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper is about adaptive robust CKF implementation on the underwater navigation system. The subject of the paper makes technically sound, but the following comment must be considered.
1-The figure (1) contains some unknown markets which must be polished from them.
2-The results figures are not clear and the readers can not justify the results very well.
3- Why the navigation time in the experiment is very short, while high-end sensors are employed for experiment analysis?
4- The high-end INS and DVL sensors are employed for the experiment, despite why the AHRS (HMR3000) which is a low-cost sensor is also invoked for this purpose?
5- The prediction covariance matrix must be presented for justification of filter performance in simulation and experiment.
6- Figure (10) has not the same format as other figures.
7- There are a variety of types of adaptive robust CKF filters in the literature. To conclude the performance of the presented method in the paper, the results must be compared with them in the experiment and simulation section.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The main previous comments of the reviewer have not been responded to very well. Therefore the reviewer emphasizes them again.
1- With respect to the reviewer's previous comment 4, the expression " Thus, the dead reckoning (DR) composed of Doppler Velocity Log (DVL) and magnetic compass are often combined with INS " is not true for cases of High-end INS utilization. As has been stated in [4], " though high-end INS provides the accuracy of the order of 0.1 deg dynamic
accuracy, their high cost sometimes urges the user to use
low-cost MEMS AHRS. Therefore, the previous reviewer's query is still remaining.
2- With respect to the reviewer's previous comment 7, the presented method in the paper is a type of adaptive robust CKF filter. In the literature there exist types of variety of this filter type. It is required to compare the presented method by one of them.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
All queries are replied to appropriately and the paper can be published in this format.
