Fault-Tolerant Control of Skid Steering Vehicles Based on Meta-Reinforcement Learning with Situation Embedding

Round 1

Reviewer 1 Report

The paper is very well written and deserves to be accepted for publication. Only a minor revision in the literature review is suggested: namely, in the introduction related how the proposed meta RL based approach refers to the similar approaches in [A]. 

[A] Discrete-Time Selfish Routing Converging to the Wardrop Equilibrium | IEEE Journals & Magazine | IEEE Xplore

Also extensive review of the English language is recommended.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

the authors present a method based on meta-RL used FTC in different fault situations, the results are validated by means of simulation.

1-What are the bounds on the magnitudes that actions can take?

2-How can the authors guarantee that the reward function R shown in (4) does not lead to a suboptimal policy action?

3-the authors should detail more and discuss the results obtained and compare them with other works

4-In the conclusions, the authors should discuss the main contributions of the article.

5-the text requires minor spelling corrections

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper proposes meta-reinforcement learning for fault-tolerant control of skid steering vehicles.  The system maintains the desired tracking performance under different types of fault situations. The paper is well organized and well-written. However, several minor points need to be revised as follows:

1.Some references that have been published by MDPI should be added.

2.Fig. 3 should be separated into Fig. 3(a) offline stage, and Fig. 3(b) online stage.
3.Algorithm 3, step 4 should be replaced by a sentence but not a symbol "<--"

4.Fig. 4(a)(b) and Fig. 5 are not clear enough, and a revision is required.

5.References need to be revised as follows:

         
Reference 1 removes GPMC, Reference 2 removes AIM, 

         
Reference 4 removes ICMA, Reference 17 removes CCC,

         
Reference 18 changes to "Remote Sensing", Reference 26 removes IROS.

6.The authors should provide a block diagram of the possible implemented system, which includes: hardware circuits, sensors, A/D converters, D/A converters, blocks of the DSP or Microprocessor software programs, sampling time of the control loops, and also the detailed descriptions of the implemented system. By adding this information, the quality of the paper can be improved.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf