A Single-Tree Point Cloud Completion Approach of Feature Fusion for Agricultural Robots

Round 1

Reviewer 1 Report

General comments: in order to improve the quality of the paper, the authors could clarify better some topics as described below. Also, should be interesting to discuss how those results could contribute to agriculture and the benefits to use those methods in the sector involved on subject of this paper.

line 21: "dataset" - review for all text

line 22: clarify "the optimal model"

line 23: means of CD and EMD?

Keywords: avoid using the same words from title

Introduction: you could explain better the importance of completing point cloud of your target (tree)

line 44: means of "DBH"

line 49: means of "LiDAR" - adopt over the text the same term (lidar or LiDAR)

line 62: in the condition of multiple pulses it is possible to detect canopy

line 75: means of "AI"

line 84: specify global and local features

Figure 1: is it required to clean the input point cloud before extraction of features?

line 305: describe better the areas (coordinates per location) and specifications of laser data acquisition (frequency of laser, etc.)

line 310: it isn't clear if you user or how you used the destructive measurements to develop your method

line 315: 30,000

line 316: 2,000

line 319-320: clarify the reason to use those number per parameter

Figure 7: "Ours" means your method? If possible, create a name or an ID

Conclusion: focus on describing it according to your main objective

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents an approach for the point cloud completion of single woods based on feature fusion. The topics of the paper are interesting. However, the following points need to be addressed to improve the quality of the work.

·         The main contributions of the work are not clear. The authors should better highlight which are the novelties and the originality of the paper with respect to the present literature.

·         The definitions of CD index, EMD index and F accuracy should be given in the manuscript.

·         What is the computational time of the proposed approach with respect to compared approaches?

·         It would be interesting to discuss how the proposed approach could work on different types of trees as well as on rows of plants, such as in vineyards or apple orchards, since only one example (one tree) is given.

·         It is not clear how the ground truth has been obtained.

·         The quality of the figures and the overall style of the manuscript should be improved.

·         No discussion about the use of the proposed approach on onboard embedded computers for agricultural mobile robots can be found in the paper.

·         The state-of-the art section should be improved by considering additional works on the topic of tree reconstruction using agricultural mobile robots. Some suggested references are the following:

 o   Bietresato, M., et al. (2016). A tracked mobile robotic lab for monitoring the plants volume and health. In 2016 12th IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications (MESA) (pp. 1-6). IEEE.

o   Ristorto, G., et al. (2017). A mobile laboratory for orchard health status monitoring in precision farming. Chemical engineering transactions, 58, 661-666.

o   da Silva, D. Q., et al. (2021). Visible and thermal image-based trunk detection with deep learning for forestry mobile robotics. Journal of imaging, 7(9), 176.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The paper has been improved as suggested and it can be accepted for publication.