Multiple Target Tracking Based on Multiple Hypotheses Tracking and Modified Ensemble Kalman Filter in Multi-Sensor Fusion
Round 1
Reviewer 1 Report
In this paper a modified ensemble Kalman filter (EnKF) is presented to substitute the traditional Kalman filter (KF) in the multiple hypotheses tracking (MHT) to deal with the high nonlinearity that always shows up in multiple target tracking (MTT) problems. A state of art is presented in the introductory part. Theoretical and simulation results are presented in the paper, showing also how the proposed method compare with other existing methods. Paper is quite well written in good English. However, here are he main points that, I consider that should be addressed further in the paper:
1) Simulations are for targets at lower sped. A discussion of how higher speed affects the proposed method will be useful.
2) Second, there should be added some discussion regarding computational time needed for the execution of the algorithm. Is this real time computation? If targets move at a higher speed, execution computational time will influence the performance of the algorithm
3) I will propose, if possible, to apply the algorithm on real targets, not just simulations and put in the paper the obtained results.
4) there are copyright problems that do not come from the articles of the authors. This should be repaired before the article to be accepted. I attached the copyright report
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
In this paper, the authors applied ensemble Kalman filter in Multiple hypotheses tracking problem.
The main motivation of the Ensemble Kalman filter is it is computationally efficient for systems with high dimensional states. In this problem, there seems to be no compelling reason to use the Ensemble Kalman filter. It is claimed that the Ensemble filter is better than EKF or particle filters. I understand that there is no theoretical guarantee that the Ensemble filter should be better.
Regarding comment 1, more extensive simulation should be done to prove that the Ensemble filter is better. I believe the simulation results tend to depends on simulation setting. Also it would be interesting if the authors provide the estimation performance of the filters (EnKF, KF, EKF, PF) when the perfect association is possible to see the underlying estimation performance of several filters.
It apparently looks like that KF shows the worst tracking performance in Fig. 9; however, it is not the case in Table 1. (Maybe I am wrong.)
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
In this paper, the author proposed an interesting approach named MHT-EnKF that is able to deal with nonlinear MTT problems in MSF scenario which introduces MHT to help the EnKF to cope with the multi-target tracking problem. However, there are still some concerns, listed as follows to improve the paper, necessary to be addressed.
1. There are some details that need to be improved, for example: On page 3, the ‘k’ in ‘scan k’ should be ‘k’. Please check the whole article.
2. There are some problems with numbering formulas. For Eq. 16, the modified equation has the same number with the original equation while this is not the case with Eq. 17, and so on.
3. On page 5, if dj(tk) is the observation vector at time ?? of scan k, then what about d(tk)?
4. What do you mean “when the model parameter state vector ??′ contains the predicted values for the observation data.”? Please clarify this point.
5. How do you set the initial value like velocity and acceleration?
6. Please explain Figure 12 further. It difficult to understand what it was trying to say.
7. In order to make this study more convecing, it is suggested to compare the proposed method with the related works done in recent one or two years? If current related methods are enough for the comparison, please give the explantions.
8. It seems from your experiments that observations can be corrected no matter how confused they are, in this case, what's the point of these observations?
9. There are some other methods in multi-sensor fusion, like
Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy, Information Fusion, 2019, 46(2019): 23-32.
Sensor dynamic reliability evaluation based on evidence theory and intuitionistic fuzzy sets, Applied Intelligence, 2018, 48(11): 3950-3962.
It is suggested to discuss the difference between the proposed method and the above methods in the Introduction section.
In short, considering the contribution of this paper, I recommend to accept this paper after major revision.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
In regard to copyright the present paper improved. However, none of the three comments made at the first version of the paper are well addressed. They are postponed for the next article, follow up of the current paper. I would recommend the third point to be addressed, namely:
3) I will propose to apply the algorithm on real targets, not just simulations and put in the paper the obtained results.
before article is accepted.
Author Response
First of all, the authors would like to take this opportunity to sincerely thank the Editor and reviewers for their insightful and valuable comments on our manuscript.
As indicated in the following responses, we have revised our manuscript to address all of the issues raised in the review process. We believe that the quality of the work has been highly improved after the revision.
Point 1: I will propose, if possible, to apply the algorithm on real targets, not just simulations and put in the paper the obtained results.
Response 1: We added a tracking result of a drone.
Reviewer 2 Report
I think the authors made an effort to improve the papers according to suggestions of reviewers' comments. I am satisfied with the revision and there are no further comments.
Author Response
Thanks for the work of the reviewer.
Reviewer 3 Report
The authors have addressed all the comments I presented before.
By considering the academic contribution, I recommend to accept this work at its current form.
Author Response
Thanks for the work of the reviewer.
