Active SLAM for Autonomous Underwater Exploration
Round 1
Reviewer 1 Report
In this paper, the authors propose another approach of Active SLAM for autonomous underwater exploration. This approach is an extension of a previous work [3]. The paper is interesting but suffer by comparison only with the odometry reconstruction. I have some comments
1) In the introduction, the lack of the other methods are not very clearly explain and when we read this part we can think about a new "best" method to perform a better pose estimation. Can you be more clear about the objectives and the lack of the other methods?
2) The part Related works cited a lot of work, but here also the objective of the paper is not clearly appear.
3) In part Methodology, the paper is based on citation [3] and [7], but here the reader needs a paper quite self contain. In this paper, we need a draw of the robot with all defined pose an angles.
4) In Results part the active slam is comparable to the dead reckoning filter. Can you compare the result of the algorithm to your previous works [3]. Because here you show that you are able to to better than odometry and it is not really challenging for your works.
5) The Results part can be improved with more simulation tests. With this test, you can show more statistic results and know the limits of the algorithm.
6) The real tests show that the algorithm works in real environment, but here it is compared only to the odometry not to another SLAM algorithm.
Author Response
We want to thank both the reviewers and the editor to carefully read our article. We appreciate their comments and suggestions and we hope that with this letter (see attach) we can clarify those parts that were not clear enough in the text.
In the attached letter we address the comments from the reviewers and we detail the actions taken to improve the manuscript. Wherever possible, ​the manuscript modifications have also been copied into this letter.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The article is interesting, generally well written and well explained.
The references are relevant and the context is well established.
The introduction sometimes gives a little too much detail, which leads the reader to prematurely ask questions (which are dealt with in the other sections) and disturbs the fluidity of reading. Adjusting the level of detail would improve the clarity of the article.
Experiences and results are relevant.
1. Introduction
[l31-33] The way of writing the sentence makes it unclear. [l37] Define "active SLAM" and why not add some references? [l43-44] Relate the notions of map entropy reduction to the exploration of new areas and robot state entropy reduction to loop closure to improve clarity [l51-52] review the English of the end of the sentence [l52-53] Redundant with L42-44 [l55] "should" seems more appropriate than "could" [l56-60] Why quality is much better? In what sense (better precision?) ? [l89-90] "The closed-form [...] propagation is also presented [...]"2. Related Work
[l105] "their framework" perhaps instead of "this framework" [l118] Advantages or drawbacks of using a pose-graph SLAM instead a particle filter (summarize the motivations in a few words). [l155-158] A bit too detailed for the intro3. Methodology
[l175] Is the distance to be covered taken into account in the choices (in addition to the other two criteria) [l178] "[...] information gain, the uncertainty [...]" [l196] Accuracy of the pressure sensors (as you specified that of the AHRS) [l208] What about delta z? [l216] SE2?4. Results
[l272] "[...] exploring a scene." [l289] New line for "Using the Gazebo [...]" [l296] "[...] performing an autonomous [...]" [l299-300 + F10] It is difficult to see on the figure: put the two images bigger and put the same colors for the same clouds for easier comparison. [F7] Increase the size of Figure 7 to make it more readable [l301] "Active SLAM" [F8] Are the views numbers related to Figure 7? If yes, indicate them in Figure 7. [l306-312] Does the number of additional scans increase linearly at distance? Can other factors be taken into account (shape of the zone)?5. Conclusion
[l330] "[...] is no a priori map."Abbreviations
[l370] Go to new line for "NBV [...]"
Author Response
We want to thank both the reviewers and the editor to carefully read our article. We appreciate their comments and suggestions and we hope that with this letter (see attach) we can clarify those parts that were not clear enough in the text.
In the attached letter we address the comments from the reviewers and we detail the actions taken to improve the manuscript. Wherever possible, ​the manuscript modifications have also been copied into this letter.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper proposes an exploration framework, based on an active simultaneous localization and mapping (SLAM) strategy, that combines three main elements, namely, a view planner, an ICP-based pose-graph SLAM algorithm, and an action selection mechanism that makes use of the joint map and state entropy reduction. Results from several tests conducted with the Girona 500 AUV, both in simulation and in the real world, are presented to demonstrate the benefits of the active SLAM strategy.
The topic is timely, and the approach is very novel. It is probably the first work in the underwater domain that combines localization, mapping, and exploration. Moreover, the paper is very well-written and structured, and is easy to follow. The results presented are very promising.
I recommend that the paper be accepted for publication.
Author Response
We want to thank both the reviewers and the editor to carefully read our article. We appreciate their comments and suggestions and we hope that with this letter (see attach) we can clarify those parts that were not clear enough in the text.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I thank the authors for the new informations in the letter and in the paper.
With the acces of the previous paper, this one is more clear.
