Robotics in Forest Inventories: SPOT’s First Steps

Round 1

Reviewer 1 Report

The article presents in an interesting way the application of a walking SPOT robot connected to a lidar for archiving a stand of trees. The article would gain completeness if an analysis of the robot's behavior depending on the type of substrate in the forest was made, the selected test area seems to be devoid of elements that hinder measurement in practice like branches, bushes, tall grasses, etc. The authors mention this in the context of further research.

Comments on the figures: I suggest standardizing the way of referring to the graphics in the figure, e.g. only using capital letters instead of the description, e.g. Upper right.

Figure 6. - line 282 insert a period at the end of the sequence.

Author Response

The article presents in an interesting way the application of a walking SPOT robot connected to a lidar for archiving a stand of trees. The article would gain completeness if an analysis of the robot's behavior depending on the type of substrate in the forest was made, the selected test area seems to be devoid of elements that hinder measurement in practice like branches, bushes, tall grasses, etc. The authors mention this in the context of further research.

As it is stated in lines 74, 312-313 and 344-345, we selected for this first test a forest stand in fairly easy conditions. As we underline at 364-367 further research activities should be carried out in more complex conditions.

Comments on the figures: I suggest standardizing the way of referring to the graphics in the figure, e.g. only using capital letters instead of the description, e.g. Upper right.

Thanks for the suggestion, we have modified Figure 2 by adding capital letters instead of the position description.

Figure 6. - line 282 insert a period at the end of the sequence.

Done

Reviewer 2 Report

In this paper authors are oresenting a Boston Dynamics robotic platform SPOT in the forestry. Although the paper is interesting to read, I don’t see any scientific contribution that is stated or presented. Therefore, in my opinion, this paper should be threated as a technical report and not scientific article.

Sections 2.3 and 2.4 should probably be the most interesting to the reader but they are written in narrative way and are kacking clarity of presentation. Additional figure with procedure overview, diagram with procedure steps would be helpful. More clear explanation of procedure end (lines 214 and 215) should be done. 
Also, in section 2.6, lines 222 and 223 (manual co-registration) should be more thoroughly explained.

RMSE results seems quite high in Table 2. Also, in Table 3, bias X values are much higher than bias Y but authors don’t comment that. There is no comment/explanation why the resulting robot trajectories are much different than planned ones (figure 3).

Finally, in discussion/conclusion part one expects some final thoughts on applicability of presented syste, comparison with current standard approaches in forestry applications.

Some corrections, mostly regarding stylle should be done. For example, sentence in line 47.

Author Response

In this paper authors are oresenting a Boston Dynamics robotic platform SPOT in the forestry. Although the paper is interesting to read, I don’t see any scientific contribution that is stated or presented. Therefore, in my opinion, this paper should be threated as a technical report and not scientific article.

From our point of view, this contribution is a scientific article. In the paper, we presented the SPOT legged robot for the first time for forestry applications and applied it to a practical case, comparing the results (DBH and location) in terms of error with a traditional surveying method.

Sections 2.3 and 2.4 should probably be the most interesting to the reader but they are written in narrative way and are kacking clarity of presentation. Additional figure with procedure overview, diagram with procedure steps would be helpful. More clear explanation of procedure end (lines 214 and 215) should be done.

Thank you, we explained more carefully the end of the tree detection procedure. In particular, the procedure presented led to the identification of additional trees outside the test area, which were then removed from the analysis (lines 219-221).

Also, in section 2.6, lines 222 and 223 (manual co-registration) should be more thoroughly explained.

We explained in more detail what we did to co-register the data. In particular, the raw data had relative metric coordinates centered at coordinate 0, 0. So, we had to transform each survey to absolute coordinate and then matching the surveyed trees with the reference one (lines 228-231).

RMSE results seems quite high in Table 2.

We also presented these results in the discussions (lines 329-331).

Also, in Table 3, bias X values are much higher than bias Y but authors don’t comment that.

We added a discussion to this aspect. In particular, given the conformation of the test area elongated in a north-south direction, survey protocols showed minor bias in identifying the position of trees in y-coordinate (lines 336-337).

There is no comment/explanation why the resulting robot trajectories are much different than planned ones (figure 3).

In the semi-natural forest environment, it is not possible to follow the trajectories hypothesized in the office because of the presence of stems, dead wood on the ground, renovation, etc. We stated it in caption of figure 4 (lines 258-260).

Finally, in discussion/conclusion part one expects some final thoughts on applicability of presented syste, comparison with current standard approaches in forestry applications.

This article presents the first forestry application of SPOT. We have tried to preset the critical issues encountered and the opportunities that these instruments will provide in the future. The comparison with current standard approaches are now more clearly highlighted at lines 349-355.

Comments on the Quality of English Language: Some corrections, mostly regarding stylle should be done. For example, sentence in line 47.

Thank you, we checked carefully the manuscript.

Round 2

Reviewer 2 Report

Authors slightly corrected previous version of the paper. However, main objections are generally unchanged. I don’t see enough scientific contribution for a journal paper. Explanation of planned path deviations presented in figure 3 are not that persuasive. Observing the maps, it seems that obstacles (trees) could be avoided in a way that will result with the trajectories that will be much more correlated to the planned ones than the paths that were obtained in the experiment. Since this is quite important for the outcome of the measurements this is an important issue.

Finally, although comments on the pros and cons of let’s call it the SPOT approach, I don’t see any suggested scenario where it could be succesfully applied.

English should be improved. There are some errors ( lines 371, 383, 411).

Author Response

Response to reviewers

Authors slightly corrected previous version of the paper. However, main objections are generally unchanged.

I don’t see enough scientific contribution for a journal paper.

Your comment is appreciated, but we firmly believe that the juxtaposition of cutting-edge platforms and their accompanying sensors against classic, tried-and-true methods could be of interest to a scientific article. One needs only consider the extensive number of papers that have evaluated the performance of new sensors compared to similar sensors of previous generations (e.g., Sentinel-2 vs. Landsat, GEDI vs. ICESat) or the abundance of research exploring the pioneering application of classification algorithms within previously unexplored domains.

Explanation of planned path deviations presented in figure 3 are not that persuasive. Observing the maps, it seems that obstacles (trees) could be avoided in a way that will result with the trajectories that will be much more correlated to the planned ones than the paths that were obtained in the experiment. Since this is quite important for the outcome of the measurements this is an important issue.

Thank you for your thoughtful observation. It appears there might be a slight misinterpretation of Figure 3. The upper part of the figure illustrates the intended path of the robot and doesn't take into account real-world field conditions. The circular plots at the figure’s top aren't indicative of the actual horizontal structure of the forest stand but rather serve as a guide to help readers visualize the 'virtual' or planned route. Despite our best efforts to adhere closely to the planned trajectory, natural obstacles of field conditions sometimes made it challenging to follow the prescribed routes. However, we clarify in the figure caption and in the text (lines 190-191) that the planned path is just a reference to test different routs lengths and survey durations.

While figure 4 shows the real situation. In particular, the first “reference” box makes it possible to identify the horizontal distribution and size of trees in the analysis area, measured by total station. In the following boxes are reported the path followed by Spot (now more visible through a marked line) together with the trees identified in each protocol.

Finally, although comments on the pros and cons of let’s call it the SPOT approach, I don’t see any suggested scenario where it could be successfully applied.

Thank you for your comment, we add in the text what we think could be the possible and best scenarios for the robot’s applications. At the current state, monitoring in the forest environment could certainly be developed in plantations, as reported in the lines 367-371.

For more traditional forest surveys, we are confident that such ground robots will be even more improved in the future (377-379), not only for structural surveys, but also to investigate specific components of the forest (such as forest-floor biodiversity – line 371-374).

English should be improved. There are some errors (lines 371, 383, 411).

Thank you, we modified the sentence at line 371. The other lines (383 and 411) refer to the references that we checked and corrected where appropriate.