Evaluation of Grain Quality-Based Simulated Selective Harvest Performed by an Autonomous Agricultural Robot

Round 1

Reviewer 1 Report

I would like to thank the authors for the revision of their work and for addressing my comments. The paper has improved a lot and my original comments have been adequately addressed. The paper reads very well, providing a fair potential contribution in the field.

Reviewer 2 Report

In my opinion, the proposed paper is well written in all its parts, with only minor flaws that can be summarized as follows:

• The citation should be written as can be found in the template (e.g., [1]);
• There is a general confusion with the unit of measurement: authors usually use Mg, but rarely "t" is used instead (e.g., 172). I would also suggest changing the data of Table 2 in "h" since it is the unit that is generally used in the rest of the manuscript;

Finally, it would be interesting to know, analytically, when SH is profitable, based on the grain prices and the grain distribution. Maybe it is a possible future work?

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The manuscript entitled “Evaluation of grain quality-based simulated selective harvest performed by an autonomous agricultural robot” moves towards the new era of precision agriculture activities and fosters the use of autonomous vehicles to the selective harvest methods. The selective harvest study focuses on the Danish context and 20 winter wheat fields are examined using 4 hypothetical scenarios. Protein content is influenced by nitrogen fertilization and cultivars, the harvest year and environmental factors and the mycotoxin infection/ grain moisture. These factors affect the quality of the crop but also affect variation within the field. The authors, distinguish between high quality areas that produce premium grains (above 10% of protein) and other areas that produce grains with 9% of proteins.

On overall the manuscript is well written but some clarifications are necessary for better presenting the results and the added value of the manuscript. The major issue is that on overall the use of SH for the wheat produced in Denmark for animal fodder is not a viable/sustainable solution (especially from an economic viewpoint), and as a result, the impact of SH cannot be highlighted through this example.

Major comments

I believe that the manuscript should address sustainability issues (economic, environmental and social) in selective harvesting in order to have a more holistic approach. This would also strengthen the theoretical part of the manuscript.

I assume that the HQ areas were manually designed within the 20 fields. I believe that the HQ areas should be more carefully selected for the simulations. For example sometimes at Table 1 the HQ areas have significant differences (%)  and this could affect the results (especially the economic viability of the selected scenario): (i) for L1 SE is 1.61, for SI is 1.72 and for TI is 1.69  and (ii) for L6 SE and SI are very close (2.20 and 2.23 respectively) while TI is 1.92.

At table 2, I cannot understand why the TI scenario has less harvest time than the SI scenario at the L1 and the M10 field. Could you please provide the calculations and the layouts of these fields? I also suggest that you provide an example in your manuscript that showcases the harvest time calculations in a single field for the 4 scenarios (C, SE, SI, TI). This could also clarify if the selection of the regions is critical to the harvest time calculation.

The HR are clear and have a small impact on the results (this is a weak point of this analysis as on overall the use of SH for the wheat produced in Denmark for animal fodder is not a viable solution), but the HC are not clearly presented. In the aforementioned example, I believe that you could provide the cost calculations for every scenario in a selected field. Do the cost calculations for the HQ produce consider the total time traveled inside the non-HQ areas and the transportation time from the field to the storage positions? Do the cost calculations include anything else except from the operational costs per hour of the autonomous vehicle (eg storage for the different grain quality types).

Minor comments

The route planning software is only discussed in the conclusions section. I suggest that you provide some information inside the manuscript.

At page 7, just before the results, the ΔHC + ΔHR / Af type is correct?  Maybe a parenthesis is needed.

Reviewer 2 Report

The paper presents a study on selective harvesting using an agricultural autonomous robot. For this purpose, a set of fields with hypothetical cereal quality scenarios were studied using the route planner of the agricultural robot Robotti. The results showed important differences between conventional and selective harvest

The work presented is interesting and the obtained results are promising. In my opinion presented methodology is clear and understandable and the paper is well written and documented.

I have no other comments.

Reviewer 3 Report

The paper investigates the route planning for an autonomous robot and two grain carts during a selective harvesting operation of grain while exploiting protein content variability on the field. An economic cost-benefit analysis is performed as well to assess the feasibility of selective harvesting under Danish conditions.

The paper is very well written. The introduction clearly and properly identifies the problem to be addressed in this paper and positions the current work and contribution in the existing body of knowledge on this topic. On most aspects the methodology is clearly documented, scenarios have been properly derived and described, results are clear and convincing and conclusions are supported by sufficient evidence. So, all in all I would rate this paper as being good to very good.

Yet, in a way the material presented in this paper does not really satisfy my engineering science appetite. Let me explain why.

First of all, I do miss a clear hypothesis and consequently I do miss a crisp and clear conclusion. A hypothesis might be something along the lines like: “Selective harvesting based on protein content variability in a field is economically feasible under Danish circumstances”. The conclusion might be something along the lines like: “Given the assumptions made, the hypothesis is very likely to be rejected on good grounds.” The authors can easily fill in the details themselves. Currently the conclusion is not more than a summary and repetition of the main findings and largely leaves it to the reader to decide what these numbers actually mean. I don’t find this to be very convincing. Taking a clearer position might also stimulate scientific progress in this domain. And this relates to a second reflection. As it currently is, I am left with the impression like “Authors properly investigated the economic feasibility of selective harvesting under very specific Danish conditions, and it does not seem to work out”. Ok, that’s a pity for the Danish farmers maybe, but honestly not very interesting nor relevant to me. I have read so many reports on studies that show that some technology or approach does not work or is not (economically) feasible under certain, mostly current and very limited, circumstances. It is not that I am against publishing negative results, on the contrary. Yet, however correct the findings in this paper are, the scientific contribution is marginal because it does not really provide much scientific insights. I wondered, what more could we learn from this study or these kinds of studies? Because there is so much to learn from these studies. Under what conditions might this concept work? Which ones of the quite broad range of assumptions used in this study is defining the results most? What needs to happen to make selective harvesting a feasible option in the future? It largely remains unclear what we really can learn from this study. And that’s a true pity I feel. It does not really push the agricultural engineering field forward. An opportunity missed. Consider this to be a heartfelt invitation.

Some details partially related with the above.

The methodology section is not very clear about the route planner used. Interestingly in the conclusions section the route planner is explicitly named as the ORP planner for the first time. Now, I don’t expect authors to describe in great detail what the inner workings of the planner are, but the references to the methodology are currently absent or hidden in some of the references used in this paper. It is a central and crucial detail of this work and needs a more explicit description. Would a different algorithm or a different parameterization of the algorithm or of the (dynamics and kinematics of the) robot yield different results and consequently interesting insights? It is largely not addressed in this work.

In my perception the discussion section essentially contains a summary and maybe a bit more detailed description of the results obtained. There is quite a lot of repetition there. For me this section should also cover two other aspects: 1) comparison to related work, 2) reflection on the assumptions made and identifying the potential impact of the assumptions on the final conclusion. There is some reference to related work. A reflection on the assumptions made is almost absent I would say. And interestingly, once assumptions are properly discussed it will be much more easy to identify and convincingly support the choice of aspects that may require further future research. Currently most items identified for future research appear by surprise and are not really based on a thorough analysis of the impact of the assumptions on the results.

As said before, the conclusion is more of a summary than a true conclusion based on a sharp research question or hypothesis.