Model Predictive Control of Quadruped Robot Based on Reinforcement Learning
Round 1
Reviewer 1 Report
The article describes the reinforcement learning tuned MPC control of a quadruped robot, the Unitree A1. The simulation environment is made up of several elements, the main ones including the PDMPC controller which is trained by a multilayer perceptron and a PPO2 RL algorithm. The control parameters of the PDMPC are adjusted by the neural network since for different motions different values should be used for adjusting the movement of the quadruped. The article is well written, but the grammar needs work. I have started to collect a few grammatical errors, but much more are present in the paper. Abstract: - Grammatical Error: Unlike the RL application of end-to-end control, our method do not need massive sampling data for training. -> Unlike the RL application of end-to-end control, our method does not need massive sampling data for training. - The presented framework provides a new choice for improve the performance of traditional control. -> The presented framework provides a new choice for improving the performance of traditional control. Introduction: - Second sentence (from line 22 through 28) has a lot of grammatical errors, and bad wording like rock goats (I guess that would be mountain goats). - Third sentence (28-30) "... inspires searchers' enthusiasm ..." -> inspires researchers (perhaps) - I feel like a resource should be given for at least the first paragraph (21-32) - 48: Grammar do not start a sentence with and - 52: first define the word the use abbreviation for PR-MPC - ... more grammatical errors present, througout whole paper Figure 6: typo: Mtalab -> Matlab - Warpper layer -> Wrapper layer 283: - "The forward velocity command gradually increase from to 1 m/s." -> Grammatical errors, from number is not defined. The article delivers interesting, but only minor findings to the community. The language quality and the technical soundness of the paper should be improved through technically involved proofreading to elevate the quality to the standards of the journal.Author Response
Dear Editor and Reviewers:
On behalf of my co-authors, we are very grateful to you for giving us an opportunity to revise our manuscript. We appreciate you very much for your positive and constructive comments on our manuscript. We have carefully studied the comments and tried our best to revise our manuscript. Please see the attachment-revised version manuscript.
The following are the responses and revisions we have made in response to your comments on an item-by-item basis.
Comment No.1: Grammatical Error in Abstract
Response 1 : All contents have been modified according to your comments.
Comment No.2: Grammatical Error in Introduction
Response 2: The first paragraph has been rewritten. all defined abbreviations are described in detail when they first appear in our manuscript. All the resource about animals come from websites. I wonder if I should add these links to the reference?
Cheetah: https://www.britannica.com/animal/cheetah-mammal
Conraua goliath: https://animaldiversity.org/accounts/Conraua_goliath/
Camel: https://animaldiversity.org/accounts/Camelus_bactrianus/
Blue sheep: https://animaldiversity.org/accounts/Pseudois_nayaur/
Comment No.3: Grammatical Error in whole paper
Response 3: The typos in Figure 6 have been replaced. The missing speed number has been added. We apologize for the poor language of our manuscript. We reviewed the manuscript several times, and worked on both language and readability. We really hope that the flow and language level have been improved.
Thanks again to the hard work of the editor and reviewer!
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The abstract, introduction and research itself are clearly presented. I like the presentation of results via pictures and graphs.
Some pictures (Figures 2, 3, 4, 6) can be improved to higher resolutions, in printed form the letters in figures are less visible
Also, I would recommend using the same font style in text and figures.
Author Response
Dear Editor and Reviewers:
On behalf of my co-authors, we are very grateful to you for giving us an opportunity to revise our manuscript. We appreciate you very much for your positive and constructive comments on our manuscript. We have carefully studied the comments and tried our best to revise our manuscript. Please see the attachment-revised version manuscript.
The following are the responses and revisions we have made in response to your comments on an item-by-item basis.
Comment No.1: Resolution of picture
Response 1: We have updated the pictures with higher resolution.
Comment No.2: Same font style in text and figures
Response 2: We apologize for the different font style in text and figures. I can't find the text font in the software where I draw pictures, and I try to use the fonts that make them look similar.
Thanks again to the hard work of the editor and reviewer!
Author Response File: Author Response.pdf
Reviewer 3 Report
The authors proposed a novel framework to integrate the advantages of model predictive control and reinforcement learning. Method do not need massive sampling data for training. Compared with the fixed parameters controller, the learned controller have better performance in command tracking and equilibrium stability.
Author Response
Dear Editor and Reviewers:
On behalf of my co-authors, we are very grateful to you for giving us an opportunity to revise our manuscript. We appreciate you very much for your positive and constructive comments on our manuscript. We have carefully studied the comments and tried our best to revise our manuscript. Please see the attachment-revised version manuscript.
Thanks again to the hard work of the editor and reviewer!
Author Response File: Author Response.pdf
