FPGA-Based Methodology for Detecting Positional Accuracy Degradation in Industrial Robots

Round 1

Reviewer 1 Report

Summary:

The paper presents an FPGA-based method of detecting deviations in robot positions using motor currents. Current signals of UR5 are analyzed using discrete wavelet transform (DWT), statistical indicators and NN classifier.

Assessment:

In general, the paper is well-written and provides solutions in robotic diagnostics and fault detection. Addressing the following suggestions and comments can help improve the paper.

General suggestions

1.      It is worth testing the performance of the system when the noise level increases. It is not clear what the exact noise level is for the results.

2.      What happens when there are multiple joint deviations? How does the ANN perform in this case?

Specifics

1.      Regarding the statement “No reported implementation of FPGA-based methods for position degradation detection” in the introduction, the authors should be specific about the FPGA- implementation they are referring to here. As far as I know, there are FPGA implementations using auxiliary sensors for detecting robot positional errors. To avoid generalization and possibly disregarding some works in literature, you could simply say the implementation is lacking…

2.      What is the proprietary feature of the FPGA design? What is the novel proprietary solution it provides? (Such as data handling or efficient framework other systems don’t provide)

3.      CO, HO abbreviations should be expanded at first usage

4.      Specify the joints corresponding to each chart in figures 12 and 13 in the figure captions

5.      It is not clear what the deviation of the calculation error means in line 519

 

Language:

1.      “Besides” is used as a subject in line 308

2.      Adjust the sentence “Then load the …” in line 330 to change it from a command

 

3.      “… each case contemplates…” in line 459 should be corrected because the cases can’t contemplate (humans can contemplate) 

 

 

Most of the paper is grammatically sound and only minor use of language issues were found. 

Author Response

I am attaching the response

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments for author File: Comments.pdf

Author Response

I am attaching the response

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors propose a new method for detecting a positional deviation in the robot’s joints by analyzing current signals from a UR5 robot using discrete wavelet transform (DWT), statistical indicators, and a neural network classifier.

1. The motivation for the work is not clear. The authors should clarify their originality (innovation).
2. In the introduction section, authors talk about some related work but omit to review fault detection in robotic systems based on deep learning methods.
3. The relative works in the following references can be mentioned in the introduction:

§  https://doi.org/10.1016/j.robot.2022.104021

§  https://doi.org/10.3390/app10082699

§  https://doi.org/10.1016/j.mechmachtheory.2023.105390

4. There are many symbols and abbreviations. A list (Nomenclature) should be given.
5. In Section 2.3, Statistical Indicators, despite all existing features, authors choose to extract only two, RMS and variance. Why this choice?
6. In Section 2.4, Artificial Neural Network, authors omit to talk about the effective composition of the proposed neural network.
7. In Figure 4, the general diagram of the proposed methodology, the proposed method seems to have two branches after the preprocessing step. Can you please comment in the 2.5. Proposed Methodology section?
8. Plotting used signals can help us understand the nature of them and thus choose adapted processing methods. Could you please show the data structure and the form of the signal samples over the entire process?
9. In Section 3: Results and Discussion, authors plot the confusion matrix, which is a good tool to assess the performance of a classification problem, but adding a table that summarizes the achieved accuracies can add more readability to the results discussion.
10. Accurate detection of positional accuracy degradation requires precise calibration and an accurate assessment of the robot's performance. Ensuring the calibration process is robust, reliable, and suitable for implementation on an FPGA can be complex. Can the authors give more details about this point?
11. Industrial robots operate in environments where noise and disturbances are common. Extracting meaningful information from sensor data while filtering out noise and unwanted signals is crucial. Can the authors give more details about implementing efficient signal processing techniques on an FPGA to handle noise?
12. The authors should conduct a comparative analysis and explain the advantages of the proposed algorithms compared with the existing ones.

Author Response

I am attaching the response

Author Response File: Author Response.pdf