FIImap: Fast Incremental Inflate Mapping for Autonomous MAV Navigation
Round 1
Reviewer 1 Report
As innovations in self-driving vehicles emerge, it's clear that robust GIS car systems are the basis for successful navigation. This article is well articulated. But, there are significant grammatical errors that need to be improved.
Overall paper is ok but it needs extensive proofreading. The overall goal of this work is to navigate the vehicle autonomously to its destination. When IMU is added with GPS it improves the accuracy in localization. The designed car successfully interfaced with an open-source OSRM map through GPRS. This makes it easy for the user to feed the path to the robot.
1. Bayes algorithm must be typed as “Bayes” not “bayes”.
2. Personal nouns such as “I, We, Us” must be avoided. In this manuscript onPage # 1, line #9, 11, page# 3, Line # 100, 101, 103, and so on, this personal noun is used frequently. Please replace it with third person personal noun.
3. There are so many abbreviations and acronym, please provide there full word or phrase once in the manuscript. (UAV, KD-Tree, LIDARs, GNSS, BFS algorithm.
Author Response
Response to Reviewer1 Comments
Point 1: Bayes algorithm must be typed as “Bayes” not “bayes”.
Response 1: We've corrected the spelling errors in the latest manuscript. And they are marked up using the “Track Changes” function
Point 2: Personal nouns such as “I, We, Us” must be avoided. In this manuscript on Page # 1, line #9, 11, page# 3, Line # 100, 101, 103, and so on, this personal noun is used frequently. Please replace it with third person personal noun.
Response 2: We've replaced all the personal nouns with third-person personal nouns in the latest manuscript. And they are marked up using the “Track Changes” function
Point 3: There are so many abbreviations and acronym, please provide there full word or phrase once in the manuscript. (UAV, KD-Tree, LIDARs, GNSS, BFS algorithm.
Response 3: We've provided the full word and phrase at the place of the first appearance of these abbreviations. And they are marked up using the “Track Changes” function
Reviewer 2 Report
The authors made a clear explanation from the previous one.
1. The authors explain that "The time complexity depends on the number of voxels that change in space". Please explain more about this, and is there any method to calculate that?
2. I can see that the author's method has better performance (Figures 5 and 6), however, the explanation about how the FIImap has more time-consuming rather than the ego-planner is limited. I encourage the authors to put some detail regarding this point.
Author Response
Response to Reviewer2 Comments
Point 1: The authors explain that "The time complexity depends on the number of voxels that change in space". Please explain more about this, and is there any method to calculate that?
Response 1:
The workspace for drones is usually unknown and complex in advance. And there are differences in the scenes observed by the stereo under different flight trajectories. The time-consuming of the inflating method depends on the difference in stereo-visual observations of the drones, specifically the number of voxels recorded in the raise and lower arrays. In section 5.2.3, we define the environmental change score to describe environmental dynamics quantitatively. And in Figures 7(b) and 8(b), the data point distributions show typical values for the differences in the scenes in the two typical scenarios.
In the latest submission, the following paragraphs have been added to discuss this conclusion in detail.
“The time complexity of map inflating depends on the number of voxels that change in space. The map inflating creates virtual occupied areas around obstacles. When inflating the map, FIImap uses the BFS algorithm to process each voxel whose occupancy value has reversed. And the BFS algorithm for each voxel has the same operation that traverses a certain range of neighbors. Therefore, the time consumption of the algorithm depends on the environment dynamic.”
Point 2: I can see that the author's method has better performance (Figures 5 and 6), however, the explanation about how the FIImap has more time-consuming rather than the ego-planner is limited. I encourage the authors to put some detail regarding this point.
Response 2:
In fact, the algorithm proposed in this paper is incremental and the time consumption of the algorithm depends on the environmental dynamics. For average time consumption, our algorithm has a bigger advantage. However, in Figures 7 and 8, it is shown that the time consumption of FIImap is sometimes more than the ego-planner. To describe this phenomenon more clearly, we compare the time consumption using time and environmental dynamics (defined as change score) as independent variables, respectively. And we do not give the average time consumption simply.
In section 4.4, the relationship between the proposed algorithm’s time complexity and environment is discussed in detail.
In section 5.2.3, we give the detailed time consumption of the inflating. And the scene change score is defined quantitatively. The paper provides the reader with more detailed data at every moment. Readers concerned about the worst-case are able to choose the applicable algorithm according to the changes in their flights. Figure 7(a) and 7(b) describe the same experiment from two perspectives, respectively, and give the main reason for the higher time consumption of the algorithm than ego-planner is the drastic scene change.
