High-Frame-Rate Camera-Based Vibration Analysis for Health Monitoring of Industrial Robots Across Multiple Postures

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors This paper proposes an innovative vibration analysis method based on high frame rate cameras for health monitoring of industrial robots, demonstrating the significant advantages of non-contact full field measurement. By combining digital image related technologies, the paper successfully achieved vibration propagation analysis under multiple postures, avoiding the quality interference and installation limitations caused by traditional sensors. The method has high practicality and can visualize vibration distribution in real time, providing a powerful tool for early fault detection and predictive maintenance. The experimental design is rigorous, covering three typical robot postures, comprehensively evaluating the dynamic response characteristics, and highlighting the influence of posture on vibration behavior. The paper has advanced technology and clear logic, laying a solid foundation for improving the reliability of industrial robots and intelligent manufacturing applications, and has important engineering value and academic contributions.   1. The paper briefly introduces sensors and HFR video methods in related work, but can further expand the comparison between wireless sensors (such as MEMS accelerometers) and acoustic methods, especially for the analysis of the advantages and disadvantages of industrial robot scenarios. This helps readers to have a more comprehensive understanding of the uniqueness of innovative points.   2. Although the experimental section mentions robot posture and vibration settings, there is a lack of detailed explanation of environmental factors such as temperature and lighting changes, which may affect the accuracy of DIC measurements. Suggest adding a description of the standardization measures for experimental conditions to improve the reproducibility of the results.   3. The results section presents the analysis in both time and frequency domains, but the transitions between each subsection (such as 4.2 and 4.3) appear somewhat abrupt. Suggest adding bridge sections to clarify the relationship between vibration visualization, displacement signals, and transfer functions, in order to enhance overall narrative fluency.   4. Some of the terms used in the paper, such as "transfer function" and "frequency response," are slightly duplicated in context and may lead to confusion. Suggest standardizing the definition of key terms and simplifying the structure of long sentences to enhance readability and international reader friendliness.

Author Response

We sincerely thank the reviewer for the insightful and constructive feedback. The statement regarding wired sensors has been carefully revised to correctly describe the operating principles of accelerometers and strain gauges.

Our detailed responses are provided in the attached file; please kindly review them.

Sincerely,
Tuniyazi abudoureheman

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents an interesting concept for real-time vibration visualization on an industrial robot by combining sub-pixel digital image correlation (DIC) and short-time Fourier transform (STFT) methods. The idea of visualizing vibrations directly on the robot structure is innovative and relevant, especially for robotic machining and condition monitoring.

However, while the experimental implementation is promising, the paper lacks quantitative evaluation, validation, and a clearly defined scientific contribution. The text is descriptive, and the abstract and introduction contain general background statements rather than a focused technical motivation. The potential impact of the proposed system would be much stronger if the authors provided a comparison with a contact-based measurement method and related their observations to the robot’s natural frequencies.

Therefore, I recommend major revisions before the paper can be reconsidered for publication.

Reccomendations:

1. Abstract and Introduction

The abstract starts with:
“Industrial robots are increasingly used to mitigate labor shortages by automating repetitive tasks and reducing dependence on human labor.”
This statement is too general and well-known; it does not contribute to the scientific motivation of the study and does not fit in an abstract of a research paper.
→ It should be removed or replaced with a sentence that directly defines the technical problem or research gap, such as limitations of existing vibration measurement or visualization methods in industrial robotics.

The same general statement also appears in the introduction, where it again does not add scientific relevance. These sections should focus instead on:

Why vibration visualization in robots is important for precision operations (e.g., machining, assembly).

What specific technical issue the proposed method addresses (e.g., lack of spatially resolved or real-time vibration monitoring).

What is the novel contribution compared to existing DIC or FFT-based approaches.

2. Methodology

The description of the DIC + STFT process is informative but lacks quantitative or algorithmic detail (e.g., sampling frequency, computation time, latency).

The term “real-time” should be supported with numerical evidence — for instance, frame rate, processing delay, or frequency range of operation.

Please provide or reference the first natural frequency (or range) of the robot’s structure in the tested configuration.

Literature values for similar 6-axis robots (e.g., ABB IRB120, KUKA KR6) show first natural frequencies in the 10–40 Hz range.

3. Results and Validation

The presented visualizations are appealing, but there is no quantitative comparison with a conventional contact-based measurement method (e.g., accelerometer, piezo sensor).

4. Discussion and Conclusion

The current “Conclusion” section should be divided into two parts:

Discussion: interpret and compare results, relate findings to known resonance frequencies, and discuss limitations.

Conclusion: summarize main contributions, practical implications, and directions for future work (e.g., integration into feedback control or predictive maintenance).

5. Figures

Improve figure quality .

Author Response

We sincerely thank the reviewer for the insightful and constructive feedback. The statement regarding wired sensors has been carefully revised to correctly describe the operating principles of accelerometers and strain gauges.

Our detailed responses are provided in the attached file; please kindly review them.

Sincerely

Tuniyazi abudoureheman

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Authors provided a manuscript focused on the development of vision based vibration analysis method for industrial robots. Manuscript provide quite interesting research, but also it contains a lot of unanswered questions. I noticed these main issues:

1. Clear declaration of the novelty of the research. What is author's input to the field? If it is new method when it should be provided validated results. Here is provided only results obtained using proposed method, without any validation of their accuracy and reliability. Therefore, novelty and aim of the manuscript must be clarified, and they should correspond with results.
2. Literature analysis also does not provide any quantitive results, just overview of other solutions, again there are no options how to evaluate performance of the method provided by the authors.
3. Methodological part lacks a lot of details, especially about data processing. Please explain how data was processed and what theoretical measurement accuracy could be achieved using such experimental setup and processing technology. 
4. Selection of the poses is also not explained, what was the motivation and are these pose somehow specific?
5. How many postures are necessary to investigate to make confident conclusions about robot health status?
6. Also in what status was the robot during experiments (operation mode, hold/running program with pauses or others).
7. Figure 5 has low quality, some graphs are not readable.
8. In the results section, obtained vibrations amplitudes are provided with micrometric accuracy, but again not clear reliability, how many times experiments were repeated, is that data statistically processed and is it confident?  
9. Manuscript finalizes without clear achievement, authors analyzes vibrations in 3 postures after processing data develops transfer functions, but the question is what should be next? How this functions further should be used for robot health monitoring?
10. Provided conclusions are not supported by quantitive results. 

Summing up all listed issues, I conclude that in my point of view, a manuscript can not be seriously improved within typical period of 10 days, therefore my recommendation is to reject and encourage resubmission after serious modification.

 

Author Response

We sincerely thank the reviewer for the insightful and constructive feedback. The statement regarding wired sensors has been carefully revised to correctly describe the operating principles of accelerometers and strain gauges.

Our detailed responses are provided in the attached file; please kindly review them.

Sincerely

Tuniyazi abudoureheman

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The article is interesting with relevant scientific value.
There are a few aspects that need to be corrected, especially regarding the extraction of
information from existing works (references) or statements like
"Wired sensors, such as accelerometers and strain gauges, measure strain by detecting changes in resistance",
where it is known that most accelerometers are piezoelectric or capacitive.
These aspects can be corrected by re-analyzing the introductory part,
without being critical aspects.
What is important to detail, however, is the experimental methodology.
Industrial applications require metrologically valid and calibrated monitoring solutions.
Even if it is a scientific work, these aspects must be taken into account.
In my opinion, it was necessary to validate the results at least for a pair of points
by using classical transducers, namely placing an accelerometer in the direction and at the excitation point of the robot
and another in one of the monitoring points. For this pair, the transfer function is classically evaluated and
compared with that determined by the optical method presented. It is necessary to describe in detail the variation of the excitation signal. From the article it can be deduced that a swept variation (with variable frequency in steps) was initially used but a sweep/chirp variation (with continuously variable frequency) was also analyzed. Disadvantages of conventional methods were indicated at the beginning. Optical methods also have their disadvantages.
It is advisable to analyze them on the method (the influence of illumination non-uniformity, defocusing during robot complex spatial movements, dirt that partially covers the random patterns on the structure, etc.).
How sensitive is the method to these current problems in the industry? Color maps are presented regarding the deformation of the structure at different frequencies.
Are they validated by simulation or, more correctly, conventional modal analysis?
Perhaps they are comparable to those provided by the manufacturer, to avoid additional measurements.    

   

Author Response

We sincerely thank the reviewer for the insightful and constructive feedback. The statement regarding wired sensors has been carefully revised to correctly describe the operating principles of accelerometers and strain gauges.

Our detailed responses are provided in the attached file; please kindly review them.

Sincerely

Tuniyazi abudoureheman

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have corrected the issues mentioned in the previous review. The abstract and introduction are now focused on the technical problem and show clear scientific motivation. The explanation of the DIC + STFT process includes quantitative details and supports the statement about real-time operation. The results and discussion sections were expanded and structured better, including comparison with related works and identified limitations. Figures were also improved in clarity. Overall, the paper has been significantly improved and the comments were properly addressed.

I have no further comments.

 

Author Response

We sincerely thank the reviewer for the positive evaluation and for acknowledging the improvements made throughout the manuscript. We are pleased to hear that the revisions to the abstract, introduction, methodological explanations, and results/discussion sections have effectively addressed the previous concerns. We also appreciate the recognition of the enhanced figures and the overall clarity of the manuscript.

We are grateful for the reviewer’s constructive feedback during the review process, which has greatly contributed to strengthening the quality of this work.

Thank you once again for your time and consideration.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Authors significantly rearranged their manuscript, partially solving my mentioned issues and provided questions. Nevertheless, I still see a major issue, related with the experiment and reliability of results. It seems that authors performed just 1 measurement of mentioned p points.  To have a reliably results, experiments must be repeated at least 3–5 times, average values as well as standard error and standard deviation must be provided, overwise how it is possible to state that is a trend or tendency instead of the result of random circumstances and impacts.

Furthermore, the arrangement of 5 and 6 sections is quite strange, I do not see a point to provide a future work as chapter placed after conclusion section. 

Author Response

We thank the reviewer for this critical comment regarding the statistical reliability of our results. The reviewer is correct that multiple repetitions are the gold standard for establishing robust trends.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors The work is significantly improved. The additions are clear and allow understanding of the experimental processes. The limits of the method are defined and research directions are indicated, including the use of data from  calibrated acceleration sensors . The advantages of the presented method are certain but the traditional methods are not as disadvantageous as they are presented.  Especially since the presented method is not easily metrologically certifiable. The methods fusion could result in great results. I would recommend for future research the fusion of data from the presented method with measurements at several points with metrologically verified accelerometers placed on the robot. The industry places great emphasis on metrological traceability and accreditation of measurement methods, aspects often neglected in academic research. A very interesting work!  

Author Response

We appreciate the reviewer's insightful comments regarding metrological traceability and the potential of multi-sensor fusion. We agree that traditional sensor-based methods have their well-established strengths and metrological certifications

Author Response File: Author Response.pdf