Autonomous Dispatch of Mobile Robots in Manufacturing Using Convolutional Neural Networks
and Yildirim Hurmuzlu
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
Please find my uploaded review comments attached.
Comments for author File: 
Comments.pdf
Author Response
Please see the attached PDF for responses to reviewer comments!
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for Authors
The article deals with the issue of applying autonomous mobile robots in the production process. Autonomous mobile robots bring better efficiency and productivity of work. However, this process depends on the methodology used to implement the robots. This work is topical and interesting for the solution.
In the introduction, the authors described the issue and problems that are and need to be solved. Related works in this area are also listed.
The next part of the work was oriented towards a case study of the application in the assembly task in the aerospace industry to evaluate three approaches to material delivery in 60 cycles, including manual walking, manual AMR dispatch and autonomous AMR deployment.
In the implementation, they used a tested autonomous dispatching framework based on a convolutional neural network (CNN) in a controlled experimental environment.
An experimental model of the workplace was created.
The production and delivery times of the assembly process were monitored.
The evaluation results show the correctness of the proposed process and the methods used.
This work brings new knowledge and has a scientific contribution to the field of production processes supported by the implementation of autonomous mobile robots.
The authors selected the right methods for solving the defined problem and presented the results appropriately.
I did not find any serious shortcomings and errors in the work. However, I have several recommendations for improving this article.
Comments:
1. I recommend reworking the presented scenarios into pseudocode form to make it clearer.
2. The work is interesting, but I would definitely recommend a video sequence from the experiment if possible. For example, as a web link to the video of individual scenarios and experiments with an embedded description.
3. In conclusion, I recommend adding plans for future research as you continue this work.
Author Response
Please see the attached PDF for responses to reviewer comments!
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors
The work is well presented and described, with clear objectives and a realistic analysis of results. The background is supported with relevant and representative references, justifying why a context-aware autonomous dispatch system is required within the manufacturing industry. The experiments are appropriate for the exploratory experimental study. However, it contains specific limitations that affect the final measurable results.
Suggestions for authors:
- The system is described as context-aware, but the logic is essentially reactive to visual states and no specification of the awareness depth is included. There is an explanation gap regarding how the system would handle complex non-sequential events, or how those non-sequential events would affect the system performance, for example, an operator needing a tool change mid-process or a fastener malfunction, which are common in the cited aerospace assembly context.
- In section 3.3, authors claim the AMR issued auditory commands to guide operators. However, it is not explained how the systems confirmed these commands were heard or followed, nor does it address how noisy manufacturing environments might affect this communication channel.
- In the discussion section, authors mention that the kickoff steps for the AMR were “not fully optimized to synchronize robot arrival with part demand”. The document lacks an explanation about how the trigger points are selected or why they were not adjusted once identified the lack of synchronization. Is there any other issue or deployment constraint that forces the solution to an unsynchronized behaviour?
- Also in the discussion section, it would be desirable to make a broader impact analysis about the benefits of deploying AMRs solutions. Not everything is about time measuring, in the aerospace industry quality is critical and AMRs outperform humans in predictability, repeatability and traceability. I encourage authors to consider including in the description other benefits of AMRs that could support the experimental study.
- The paper contains some minor grammatical oversights in technical language, failing to hyphenate compound adjectives: context-aware (l.37, l.74, l.94, l.210); real-world (l.54, l.397); python-based (l.146, l178); real-time (l.41, l.72, l.78, l.93, l.180, l.189); two-stage (l.182); CNN-based (l.79, l.92, l.123, l.137, l.274, l.376, l.387, l.392, l.414).
Author Response
Please see the attached PDF for responses to reviewer comments!
Author Response File: Author Response.pdf
