Implementing Real-Time DOV Compensation: A Practical Approach Using a MLP on an NPU Embedded in an INS

Round 1

Reviewer 1 Report

(1)  MLP model training is mentioned several times in this paper, but the author does not provide more specific explanations. We recommend that the authors provide more implementation details, including the specific architecture and parameters of the MLP. 

(2)  Compared with the existing DOV compensation method, what is the computational complexity and real-time performance of this method? We recommend that the authors compare with existing or other widely used schemes to verify the advantages and innovation of MLP in real-time DOV compensation schemes. 

(3)  Although the INS and single-antenna GPS receivers used are described in detail in this paper when evaluating real-time deflection vertical compensation performance, we believe that more details are needed, such as experimental methods and complete procedures. We would like the authors to reorganize this section and provide a more detailed explanation. 

(4) The article is clear when describing the relationship between the SRTM data in Figure 6 and the DEM, DTM, and DSM models. However, more information is needed to explain the differences between the figure and the actual application scenarios. 

 

(5)  In section 4.4, real-time evaluation of gravitational perturbations and DOV compensation is mentioned, but detailed methods, experimental results, and specific evaluation indicators are lacking. We recommend that the authors provide more details to support the content of this section.

(6)  The author contribution section needs to more clearly describe the specific contributions and experimental results. The experimental design, method, and description of results need to be further refined.

It is recommended that the authors should check the language carefully, so as to further improve the quality of the manuscript. In addition, the authors should give appropriate punctuation at the end of each formula, especially where there are several formulas in succession.

Author Response

Please see the attachment. 

Author Response File: Author Response.pdf

Reviewer 2 Report

In this paper, the authors present a real-time method for estimating gravity disturbances using a Multilayer Perceptron (MLP) neural network trained on ground surface gravity data derived from the EGM2008 gravity model. The results obtained through field experiment demonstrated the effectiveness of the authors approach in enhancing positioning accuracy.

1.      In the abstract section of the paper please reframe this statement “In this paper presents analytical findings regarding the influence of gravity disturbance on INS. A data from the high-precision gravity model, EGM2008, we introduce a novel real-time 11 gravity disturbance compensation approach for INS.”

2.      On page 4 of the manuscript, it was stated that “However, it’s essential to note that gravity potential is conventionally expressed in spherical coordinates.” For confirmation of this statement, supply the source and reference.

3.      The software package used to generate plots in Figures 4 and 5 was not stated. Please, state the software package.

4.       How does the novel real-time gravity disturbance compensation approach for INS developed compared to the existing models? 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I have no further comments！