Adaptive Path Planning for Fusing Rapidly Exploring Random Trees and Deep Reinforcement Learning in an Agriculture Dynamic Environment UAVs

Round 1

Reviewer 1 Report

The manuscript designed a solution for online adaptive 3D path planning based on the fusion of RRT and DRL for UAS. The investigated topic is interesting and the manuscript has certain contributions. The paper can be refined by considering the following comments:

 1. The literature review in Introduction can be improved. First, when introducing that Unmanned Aerial Systems (UASs) open new paths that advance the solution of existing problems, the application of UAVs/drones for package delivery in logistics can be introduced such as ‘Efficient routing for precedence-constrained package delivery for heterogeneous vehicles (2019)’. Second, after introducing two types of motion planning: global path planning and local path planning, more solutions regarding the developed path planning strategies are suggested to be analyzed such as the A* algorithm and Q-learning algorithm proposed in ‘Path Planning for Wheeled Mobile Robot in Partially Known Uneven Terrain (2022)’ for robot path planning in unknown environments. Furthermore, when introducing that there is a large volume of work dealing with path planning in the literature, the optimal control theory-based algorithm is also popular for path planning as shown in ‘An integrated multi-population genetic algorithm for multi-vehicle task assignment in a drift field (2018)’ and ‘Clustering-based algorithms for multivehicle task assignment in a time-invariant drift field (2017)’.

2. The UAS used in the title of the manuscript and in Figure 1 are suggested to be changed to UAV, which is more clear.    

3. A period or a comma is needed at the end of each equation.

4. Some relevant state-of-art path planning algorithms are suggested to be compared to verify the performance of the designed path planning algorithm, and the computational running time of the algorithms is also needed for a fair comparison. 

Author Response

Dear reviewer,

Attached is the document containing all the requested considerations.

Sincerely

Guido Szekir Berger

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors try to forcefully adapt problems solved with the autonomous movement of robots and ground vehicles to the application of unmanned aerial systems and, in this manuscript, to unmanned aerial systems used to monitor olive trees. This situation makes no sense because the manuscript does not address an essential feature of unmanned aerial systems, namely that they can move above objects. The authors need to learn about agriculture and horticulture and, in this case, olive growing.

Olive trees are not tall enough for a drone to be unable to fly above them, so it doesn't have to pick its way between the trees. If the authors don't think so, I recommend scientific articles on the use of drones for spraying fruit and olive trees.

Training facilities that are supposed to imitate olive trees resemble skyscrapers in the center of big cities in appearance, positioning, and height rather than trees - no sense.

Inspection of tree crops can be done above them, and for very thorough inspections done from the side of the trees, drones move systematically along all the paths between the trees and inspect all the trees, not just randomly selected ones. So there is nothing there for autonomous movement and path selection. In addition, the authors, in forming the issue, should have taken into account that drones have a limited time in the air and should independently return to the place of takeoff.

The authors got carried away by their imagination. I recommend that they justify well the sense of applying the proposed solutions in agriculture. The manuscript is addressed to a scientific journal - Agriculture, so it should not contain a description of solutions detached from agriculture. Justifying the topic only because unmanned aerial systems are used in agriculture to monitor crops does not sufficiently explain the further actions described in the manuscript. Mainly since the authors write in the introduction that the paper is about unmanned aerial systems used to monitor trees and the tests and further simulation studies described in the manuscript are about driving a car between trees.

The computer solutions presented in the manuscript on applying Rapidly-Exploring Random Trees (RRT) and Deep Reinforcement Learning (DRL) algorithms which are not much of a scientific novelty, especially for planning studies of the autonomous movement of robots and ground vehicles. So I rate the article, and from this point, as weak.

The authors should think about how to modify the described solutions so that they can be applied to unmanned flying equipment.

Suppose the authors still intend to publish the manuscript in the journal Agriculture. In that case, I recommend describing precisely what functions the drone is supposed to perform and what functions have been assigned to it from the point of view of agriculture and only then simulate its route. 

Author Response

Dear reviewer,

Attached is the document containing all the requested considerations.

Sincerely

Guido Szekir Berger

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper presents an adaptive path planning for UAS fusing rapidly exploring random trees & DRL in an agriculture dynamic environment. The topic is certainly worthy of investigation nowadays, but the manuscript suffers from the following shortcomings:

·         The research question and has not been put forward clearly.

·         A comparison table between the proposed system and other related studies is highly recommended to be added in Section 2 (Background and Related Works).  Then, motivations can be easily drawn.

·         It is recommended to add more references and highlight the importance of the Fourth Industrial Revolution pillars and briefly emphasis the UAS and its intelligent applications including agriculture. Here some suggested references: 

https://ieeexplore.ieee.org/document/9515736

https://www.mdpi.com/2071-1050/13/11/5908

https://link.springer.com/book/10.1007/978-981-19-2027-1

·         Will the considered UAS is an autonomous system? Cannot see how? Should be explained in depth.

·         It is clear that the proposed UAS has starting point when plaining the path. Yet ,does it have a return point? Clarify.

·         Does the altitude of the UAS has been considered in the path planning?  

·         Figures 2, 8, and 9 should be improved from size and resolution perspectives.

·         Try to avoid self-citation and reduce those to the minimum.

·         Overall, the manuscript suffers from some issues, so it is recommended to be a minor revision.

Author Response

Dear reviewer,

Attached is the document containing all the requested considerations.

Sincerely

Guido Szekir Berger

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Autorzy uwzględnili wszystkie moje uwagi do poprzedniej wersji manuskryptu. Dokonali istotnych poprawek i uzupełnień w tekście. Rękopis zyskał nowe, cenne wartości z punktu widzenia nauk rolniczych.

Nie mam dalszych uwag.