Improved Rapidly Exploring Random Tree with Bacterial Mutation and Node Deletion for Offline Path Planning of Mobile Robot

Round 1

Reviewer 1 Report

Authors presented the simulations of path planning algorithm for mobile robot. In general, the paper is well written and explained with sufficient level of details. I recommend the publication following minor corrections/clarifications.

• Authors should discuss the experimental implementation of this algorithm.
• It seems that only simulations were performed. Although this reviewer does not think it is a problem, it would be beneficial if the authors make it more clear throughout the paper. 
• A deeper discussion about dynamic obstacles would be beneficial, since it is a common condition on practical applications in unstructured environments. 

Author Response

Authors’ Response to Reviews of

Improved Rapidly-exploring Random Tree with Bacterial Mutation

and Node Deletion for Offline Path Planning of Mobile Robot

Aphilak Lonklang and Janos Botzheim

Electronics, electronics-1699087

We thank the associate editor and reviewers for the valuable feedback. We have thoroughly addressed all the comments from the reviewer described in this document in detail. We have classified the main improvements into the following list:

Note: All changes are covered in blue color in the revised article.

1. Authors should discuss the experimental implementation of this algorithm.

Thank you for the comment. The implementation of the experiment was conducted in MATLAB programming language. The section was amended according to the comment.

The MATLAB programming environment was used to realize the experiments.

2. It seems that only simulations were performed. Although this reviewer does not think it is a problem, it would be beneficial if the authors make it more clear throughout the paper.

Thank You for the comment. Indeed only simulations were performed and the modifications were added according to the comment.

To test the final solution from the proposed algorithm, the simulation using Robot Operating System collaborated with MATLAB programming was conducted.

3. A deeper discussion about dynamic obstacles would be beneficial, since it is a common condition on practical applications in unstructured environments.

Thank You for the comment. Modifications were added in the Conclusion section.

As a future work our proposed algorithm can be extended to deal with more complex environments (e.g. with dynamic obstacles). In this research, the feasible mapping algorithm was used only for pre-processing. The concept of feasible mapping can be applied during each iteration by adding the possible coordinates to iterate through or deleting coordinates in the obstacle's regions. The size of the feasible mapping vector will be changed depending on the dynamic obstacles that the robot is facing.

Reviewer 2 Report

Authors proposed an improved solution (algorithm) to optimize the path planning of mobile robots in large and complex environment. Authors added the pre-processing with feasible region mapping and post-processing with Bacterial Mutation and Node Deletion operators to improve the RRT∗ algorithm.  The pre-processing algorithm allowed to reduce the computation time by reducing the number of unusable nodes. In case of the post-processing algorithm, the overall path length and number of nodes to represent a smooth path was optimized. Authors concluded that the generated path from the proposed algorithm was shorter and smoother comparing to the traditional  RRT∗ and traditional RRT∗ with post-processing algorithms.

Introduction section is enough sufficient but it could be extended by other methods applied to this matter. The experimental environment and proposed algorithms are clearly explained. Experimental results were also clearly presented. The last conclusion section supports the experimental results. The article lacks in discussion section. The problem was not enough discussed itself and with other solutions. Mostly authors focused on RRT. The discussion section should be added to this article. But generally speaking the presented solution is interesting and worth publishing.

Author Response

Authors’ Response to Reviews of

Improved Rapidly-exploring Random Tree with Bacterial Mutation

and Node Deletion for Offline Path Planning of Mobile Robot

Aphilak Lonklang and Janos Botzheim

Electronics, electronics-1699087

We thank the associate editor and reviewers for the valuable feedback. We have thoroughly addressed all the comments from the reviewer described in this document in detail. We have classified the main improvements into the following list:

Note: All changes are covered in blue color in the revised article.

1. Introduction section is enough sufficient but it could be extended by other methods applied to this matter.

Thank You for the comment.

We added more introduction about two other methods that are dealing with the lack of the exploring process as seen in the "Introduction" section.

• The Hybrid Bidirectional Rapidly-exploring Random Trees Algorithm
• The Density Avoid Sampling (DAS) Technique for RRT*

2. The article lacks in discussion section. The problem was not enough discussed itself and with other solutions. Mostly authors focused on RRT. The discussion section should be added to this article.

Thank You for the comment. The following modifications were added to the corresponding Discussion section.

A feasible region mapping algorithm was proposed to reduce the number of unusable nodes for the coordinate exploring section of the RRT* algorithm. As seen in the results, the number of unused nodes was reduced. The applied changes are leading the algorithm to explore the entire environment more efficiently. After getting the first solution from the proposed RRT* algorithm, the Bacterial Mutation and Node Deletion algorithms were applied to smoothen and minimize the path length. Comparing the proposed algorithm with other algorithms in Table 2, the proposed algorithm demands less computational power.

During the conducted experiments our proposed algorithm was compared to traditional RRT* algorithm. According to Table 2 the proposed algorithm returns a less complex path that is also smoother. During the experiments a traditional method, the A* was also compared to our proposed algorithm and the comparison resulted also in the favor of the hereby proposed method.

The results from commonly used algorithm (A*) for both the simple and the complex environment were added to "Comparing the Result with Commonly Used Algorithm" section.