Research on Path Planning and Control of Intelligent Spray Carts for Greenhouse Sprayers

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors "address the challenges of inefficient path planning, discontinuous trajectories, and in adequate safety margins in autonomous spraying vehicles for greenhouse environments".

The authors also redesigned the heuristic function of the A* algorithm aiming to optimize the node expansion efficiency. Also, "an adaptive look-ahead distance pure pursuit algorithm was designed for trajectory tracking". 

Overall the paper is well written and all the details related to the A* algorithm redesign and for the control function pursuit algorithm are clearly presented.

Unfortunately the paper still lacks some important aspects related to the following:

1. In the abstract the authors states that "the proposed approach improves planning efficiency by 38.7%" but this is the only reference in the paper for this value. How was computed? The authors should give proper details in text.
2. Similar remarks for other values mentioned in abstract section: "reduces node expansion by 16.93%, shortens the average path length by 6.3%, and decreases the path curvature variation by 65.3%".
3. No details related to Simulink-ROS just mentioned in abstract and conclusions sections.
4. In the "Real-vehicle Path Planning Experiment" section there is a lack of details related to the experiment. Are there multiple repetitions for each scenario? If it was just one simulation for each scenario the results should be the same if the simulation is made 2 or more times? Some more details should be given in order to be sure that future readers will have this information.

 

Author Response

Reviewer 1

Comments 1: 

In the abstract the authors states that "the proposed approach improves planning efficiency by 38.7%" but this is the only reference in the paper for this value. How was computed? The authors should give proper details in text.

Response 1: 

Thank you for pointing this out. We agree with this comment. By refining the heuristic function and removing redundant turning points, we performed multiple simulations on various grid maps. This allowed us to collect data on node expansions, processing time, number of turns, and path length—both before and after the algorithm optimization. From these results, we derived relevant metrics such as the efficiency improvement rate and path reduction ratio. The specific modifications are elaborated in Figure 5, Table 2, and the text highlighted in red between lines 163 and 173.

Comments 2:

Similar remarks for other values mentioned in abstract section: "reduces node expansion by 16.93%, shortens the average path length by 6.3%, and decreases the path curvature variation by 65.3%".

Response 2:

Agree. By refining the heuristic function and removing redundant turning points, we performed multiple simulations on various grid maps. This allowed us to collect data on node expansions, processing time, number of turns, and path length—both before and after the algorithm optimization. From these results, we derived relevant metrics such as the efficiency improvement rate and path reduction ratio. The specific modifications are elaborated in Figure 5, Table 2, and the text highlighted in red between lines 163 and 173.

 

Comments 3:

No details related to Simulink-ROS just mentioned in abstract and conclusions sections.

Response 3: 

Agree. In this study, the control algorithm was modeled in Simulink and compiled for deployment on Huawei's MDC intelligent driving computing platform. It was then integrated into the ROS environment to conduct communication and path-tracking experiments. Specifically, the compiled Simulink model produces executable nodes within ROS, which communicate with both the path planning module and the RVIZ visualization tool, enabling real-time comparison between the actual vehicle trajectory and the planned path. The visualization results are presented in Figure 7. Furthermore, as noted in the revised manuscript, corresponding experimental details have been added in the section marked in red between lines 369 and 372.

Comments 4:

In the "Real-vehicle Path Planning Experiment" section there is a lack of details related to the experiment. Are there multiple repetitions for each scenario? If it was just one simulation for each scenario the results should be the same if the simulation is made 2 or more times? Some more details should be given in order to be sure that future readers will have this information.

Response 4:

Thank you for pointing this out. We agree with this comment.Further experimental details are provided in the Conclusions section (as indicated in the text highlighted in red between lines 369 and 372), with additional supporting data presented in Figure 11 and Table 6. During the experimental phase, rigorous testing was conducted to ensure the reliability of the results.

 

While a single experiment conducted under ideal controlled conditions would be expected to yield consistent results upon replication, real-world physical systems are subject to inherent stochastic disturbances. These include sensor noise, slight variations in ground friction, and fluctuations in communication latency. To enhance the robustness of our conclusions and derive statistically meaningful performance metrics, each scenario described in this study was subjected to multiple independent experimental repetitions—specifically, three or more trials. The quantitative results reported herein represent the averages obtained from these repeated trials, forming a reliable basis for evaluating the system's steady-state performance. This methodology helps to mitigate the influence of random variability and provides a more accurate representation of the system’s performance under realistic operating conditions.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses the path planning and control of intelligent spray carts in greenhouse environments. The authors propose a hierarchical framework combining an improved A* algorithm for global planning, state-space sampling with multi-objective evaluation for local planning, and an adaptive pure pursuit controller for trajectory tracking. The system is validated in both simulation (ROS–Simulink) and real-vehicle experiments using a Huawei MDC300F platform.

The manuscript is generally well written, technically sound, and supported by convincing experimental results. However, the novelty is somewhat incremental, and there are areas where the presentation and discussion could be strengthened.

1. While the integration of improved A*, Bézier smoothing, and pure pursuit control is well executed, the novelty is primarily in the system integration rather than new algorithms. Please highlight this clearly in the introduction and conclusion.
2. The paper reports improvements in planning efficiency, but more detail on computation time, hardware requirements, and scalability would strengthen confidence in real-world deployability.
3. The test scenarios are strong, but more quantitative comparisons with baseline methods (beyond A*) would provide clearer benchmarking.
4. Figures are in low quality.

Comments on the Quality of English Language

The manuscript is generally understandable and logically structured. Technical terminology is appropriate, and equations and figures are well presented. However, the language could benefit from polishing.

Author Response

Reviewer 2

|Comments 1: 

While the integration of improved A*, Bézier smoothing, and pure pursuit control is well executed, the novelty is primarily in the system integration rather than new algorithms. Please highlight this clearly in the introduction and conclusion.

Response 1:

Thank you for this insightful observation. We fully agree with your comment. In the revised introduction, we will explicitly refine our framing to shift the emphasis from “proposing a novel algorithm” to “constructing and validating an efficient, application-oriented integrated system.” This will more clearly articulate that the main contribution of this work lies in the engineering implementation, system integration, and comprehensive empirical assessment of this specific technical combination.Accordingly, the conclusion will be revised to reinforce this focus, summarizing the value of the study in terms of delivering a practical, empirically validated integration scheme. We will also elaborate on its potential for real-world engineering applications and suggest directions for future optimization.We sincerely appreciate your constructive feedback, which has been invaluable in enhancing the clarity and rigor of our manuscript.

 

Comments 2:

The paper reports improvements in planning efficiency, but more detail on computation time, hardware requirements, and scalability would strengthen confidence in real-world deployability.

Response 2: 

Agree. By refining the heuristic function and removing redundant turning points, we performed multiple simulations on various grid maps. This allowed us to collect data on node expansions, processing time, number of turns, and path length—both before and after the algorithm optimization. From these results, we derived relevant metrics such as the efficiency improvement rate and path reduction ratio. The specific modifications are elaborated in Figure 5, Table 2, and the text highlighted in red between lines 163 and 173.

 

Comments 3:

The test scenarios are strong, but more quantitative comparisons with baseline methods (beyond A*) would provide clearer benchmarking.

Response 3:

We are deeply grateful to the reviewers for their thorough and constructive feedback. We are in full agreement with their evaluation and will conduct a thorough investigation of these issues in our ongoing research.

Comments 4:

Figures are in low quality.

Response 4:

We thank the reviewer for this suggestion. In response, we have improved the clarity of all images by enhancing their resolution, contrast, and labeling to ensure they meet the required standards.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper validates, through both simulation and extensive real-vehicle tests in multiple scenarios, is particularly strong and adds significant weight to the claims.

1. The introduction highlights the importance of connected and autonomous vehicles, but the specific research gap remains a bit unclear. Could the authors clarify or compare with existing works?
2. As a reviewer, I am really impressed with the real-world testing in section 5. The authors mention in the abstract and conclusion that our method improves "time and energy efficiency," but the results focus mostly on path smoothness and safety. Did the authors measure the actual battery usage or task completion time during those tests? I'm curious to see some data backing up the efficiency claims.
3. The heuristic function redesign in Equation (2) is interesting. The authors provide more intuition on how the weight coefficient β was chosen. Was it tuned empirically for the greenhouse environment, and is there a risk that this tuning might be scenario-specific?
4. While the multi-objective cost function for local planning is well-defined, the chosen weights (k_s = 0.3, k_r = 0.3, k_e = 0.4) are stated but not justified. Were these weights determined through systematic analysis or empirical observation? A brief justification would strengthen the methodology.
5. 4 contains a lot of information. Could the authors consider simplifying the visuals or providing more detailed explanations in the captions?
6. The conclusion briefly summarises the contributions. Would it be possible to include a clearer statement of the main limitations of the current study to better frame future work?

Author Response

Reviewer 3

Comments 1: 

The introduction highlights the importance of connected and autonomous vehicles, but the specific research gap remains a bit unclear. Could the authors clarify or compare with existing works?

Response 1:

Agree. We have revised the introduction to address this point. While intelligent connected vehicles generally require robust performance across diverse operational environments, with high demands on technological complexity and safety, the intelligent spraying vehicle presented in this study is designed to meet its performance metrics within a specific operational context. We thank the reviewer for this valuable insight, which directs our subsequent work toward developing algorithms for intelligent vehicles in open and unstructured environments.

 

Comments 2:

As a reviewer, I am really impressed with the real-world testing in section 5. The authors mention in the abstract and conclusion that our method improves "time and energy efficiency," but the results focus mostly on path smoothness and safety. Did the authors measure the actual battery usage or task completion time during those tests? I'm curious to see some data backing up the efficiency claims.

Response 2:

We agree with the reviewer's insightful suggestion to incorporate direct energy efficiency measurements, like battery consumption, as a key performance metric. This will be a primary focus of our subsequent research.

 

Comments 3:

The heuristic function redesign in Equation (2) is interesting. The authors provide more intuition on how the weight coefficient β was chosen. Was it tuned empirically for the greenhouse environment, and is there a risk that this tuning might be scenario-specific?]

Response 3:

Agree. We have appropriately calibrated the β value to accommodate greenhouses of different dimensions, thereby ensuring that the optimization function exhibits robust adaptability and practical utility across structures of varying sizes. It is important to note, however, that this calibration may introduce scenario-specific limitations. In future work, we will prioritize the development of path-planning algorithms tailored for open environments, in order to overcome these constraints and enhance the generality of our approach.

Comments 4:

While the multi-objective cost function for local planning is well-defined, the chosen weights (k_s = 0.3, k_r = 0.3, k_e = 0.4) are stated but not justified. Were these weights determined through systematic analysis or empirical observation? A brief justification would strengthen the methodology.

Response 4:

We thank the reviewer for this comment. The weighting ratio was determined through systematic experimental testing to identify the optimal local path.

 

Comments 5:

4 contains a lot of information. Could the authors consider simplifying the visuals or providing more detailed explanations in the captions?]

Response 5: 

Agree. The optimal controller was selected through a structured process: first, the principles of the lateral controller were defined, followed by comparative experiments involving several controllers designed for greenhouse applications. The final selection was made based on a thorough evaluation of the experimental results.

 

Comments 6:

The conclusion briefly summarises the contributions. Would it be possible to include a clearer statement of the main limitations of the current study to better frame future work?

Response 6:

We agree with this observation. The study's limitations have been acknowledged in the concluding section.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have no other comments on the paper.

Comments on the Quality of English Language

The paper needs careful language improvement.