The Design a TDCP-Smoothed GNSS/Odometer Integration Scheme with Vehicular-Motion Constraint and Robust Regression

Round 1

Reviewer 1 Report

The manuscript is overall merited. The authors extent the Kalman filter to integrate GNSS and odometer and achieved meter-level accuracy. This is a contribution of knowledge. The information in introduction and methodology parts should be better conveyed.

Detailed comments:

1. The three techniques are not proposed by the authors. They are more implementations than inventions.
2. The introduction part is overall expected to write in a more concise and logical ways. For example, in the third paragraph, a plenty of techniques have been introduced, which makes the readers to get lost. The details shall be better managed to introduce the pros and cons of the current techniques.
3. P2, Line 49. It is wrong. The maximum range error due to multipath is half chip (equivalent to about 150m for GPS L1), not one chip length.
4. P2, Line 85. Do you mean “a little influence on NLOS reception”, or “little influence on NLOS”? Explain why.
5. P2, Line 94 — P3, Line 97. It is not precise. “GNSS-friendly environment” is not an academic term. The author indicate the number of visible satellites in open environments. How many constellations have been involved? You could refer to the following reference to back your statement — less satellites would be received due to signal blockage in urban areas while more reflected signals can be received in such environments. (Gao, H., & Groves, P. D. (2018). Environmental context detection for adaptive navigation using GNSS measurements from a smartphone. Navigation: Journal of the Institute of Navigation, 65(1), 99-116.)
6. P3, Line 117. Satellite availability and spatial geometrical strength are not additional sensors.
7. Section 2. It is suggested to introduce the overall integration scheme first (Figure 2 & 3), then each element. This will make the manuscripts more accessible. Suggest to introduce more details about the mechanism of odometer. When the velocities of the wheels on two sides are not equal, such as during turning, what does the odometer measure?
8. The equation following equation (27) is wrong.
9. Section 3. Suggest to add the specification of the odometer. Suggest to add the detailed performance analysis and comparison in the GNSS-challenging scenario. (In a table form like Table 2 & 3)
10. Table 4. How did you estimate the receiver clock error? If it is obtained from wls or KF using all received satellites, it would not be accurate as some are NLOS/multipath contaminated. There are likely to be some problems with the range errors list in the table. For LOS range errors, it should be some very small values around zero, positive and negative. However, the listed range errors are all positive.

Author Response

We would like to thank you for the objective and thorough review of our paper. We have addressed your comments in the following point-by-point response. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors presented a method to counteract the limitations of GNSS for automotive navigation in challenging environment. The method is based on the integration of the odometer (which is always available in present generation cars) in an EKF completed by time differenced carrier phase measurements and WLS regression. Each item in the proposed method is introduced to counteract one or more weak points of GNSS (e.g. multi-path, low number of measurements, cycle slips, zero-velocity condition). The paper is relatively well written and the theoretical development is supported by a good experimental analysis.

However, I have a number of minor comments the authors should address before considering the paper ready for publication.

1) One point is missing in the analysis of scenario and of the state of the art. Indeed, low cost compact INS units are currently on the market which could be used to support GNSS operations for automotive applications. My personal idea is that those inertial unit are not enough accurate to be competitive against odometers, by the way the authors should comments about this point (i.e. cheap INS vs odometer) or, if possible, should add relevant experiments to their analysis.

2) please consider using forward and lateral motion/coordinates instead of along-track and cross track one throughout the manuscript

3) Define the acronym OBD. Moreover you wrote in the text both “OBDII” and “OBD II”. Please make uniform.

4) Section 2.1. The section opens with a very traditional and well-assessed definition of EKF equations. Please consider shortening the section to better highlight your approach which is instead able to include delta measurements.

5) Page 4 line 160. what is intended as earth frame herein? WGS84? Local datum? Local ENU? Please clarify.

6) Section 2.2. Same as for 4). The pseudo range (but also doppler and carrier-phase) model and the relevant discussions are well-known and well-assessed in the literature. Please consider revising the section to remove any well-known content from the literature and just better focus the manuscript on novel content.

7) Page 6, Eq.12. Theoretically, doppler shift measurements are not independent from the pseudorange ones. Hence, they should not be used at the same time within an EKF implementation. Indeed, thanks to the dynamic models of the EKF, the velocity is updated even though you just use pseudorange measurement only. Please consider removing Doppler measurements or testing about their actual utility within your EKF.

8) Page 6, Eqs. 13-17. Even those equations are pretty standard, so consider shortening this section, just discussing novel content, i.e. the integration of TDCP measurements in the EKF.

9) Page 11, figure 3. Improve the layout of the process flow (text spreading outside the boxes).

10) Page 11, line 458. It is not clear to the reader the way in which the reference solution by the tactical-grade INS is derived, and the errors of the tested methods are evaluated? Any calibration involved?

11) Page 14, line 508. So, the e-frame seems to be an East North Up frame. What's the origin of that frame?

12) Page 20, Table 5. The notation "which road" in table 5 is unclear. I suggest the authors to use different, more technical, terminology or quoting a suitable reference for that “parameter”.

Author Response

We would like to thank you for the objective and thorough review of our paper. We have addressed your comments in the following point-by-point response. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

• The results need to be presented more clearly- see example figures 3 or 8. In Figure 8 I can not read the lines indicators.
• Which type of antenna did you use in your test? 

Author Response

We would like to thank you for the objective and thorough review of our paper. We have addressed your comments in the following point-by-point response. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

The paper focus to importance design a TDCP-Smoothed GNSS/Odometer means is beneficial. Improving the paper in better describing the needs is recommended. The paper must be improved in English. Some sentences or even paragraphs are not clear and hard to read. The definition " Receiver autonomous integrity monitoring (RAIM) adopts the same working principle of consistency checking to calculate the integrity of GNSS receivers." is not tru, it necesery precisly described diferences.

Results, discussion and conclusion are mixed together. This can be acceptable, but in the paper they are mixed completely. Part of results is followed by sentence of discussion, than next results, some conclusion. The paper must have clear form and one flow.

 

Author Response

We would like to thank you for the objective and thorough review of our paper. We have addressed your comments in the following point-by-point response. Please see the attachment.

Author Response File: Author Response.pdf