Three-Dimensional Formation Control for Robot Swarms

Round 1

Reviewer 1 Report

This paper addresses a distributed 3D algorithm to coordinate a swarm of autonomous robots (ARs) to spatially self-aggregate in arbitrary ways based on local interactions. To accomplish the task the AR has local proximity sensors to measure the relative coordinates of its nearby AR. For this it was necessary to add a leader that is able to locate and communicate in global coordinate systems, while accessing the arbitrary 3D shape specified by the user. The leader sends communication signal directly to any other AR. According to the authors, the objective is to maneuver ARs while maintaining a user-specified 3D formation, so that the network connection of all ARs is maintained during the maneuver. According to the authors, AR results in 3D formation form and does not need an AR's global location, except for the leader. According to the authors, the paper is new to controlling the 3D formation with form, so that the network connection is maintained while the ARs maneuver based on local interactions. The authors assure that the proposed control provides reasonable precision in the face of significant AR glitches and motion error. Some experiments were demonstrated using MATLAB simulations, and with that the authors presented superior performance of the proposed formation controls.

 

The work is well written and readable in terms of scientific writing. Some considerations will be carried out in the present work in order to improve the quality of the work. The yellow line used in Figures 2, 4, 6 and 8 could be thinner for better visualization of the points behind. Place the dots in the figure in front of the line and reduce the thickness of the yellow line. Another suggestion is to place lines of different colors to highlight each experiment presented. For example, the line in Figure 4 could have a different color than the line in Figure 2, and so on. Another improvement is to put a table of symbols, as we have many variables throughout the work, this table of symbols would make it much easier to read the article. Another important and necessary recommendation is to create a section that presents the state of the art and differentiates the current article from other published articles in the literature. Some of the published articles also present a formalization in graphs for decision making in collective robotics, so it is important to cite the articles below, making a contrast with the present article. Because it is necessary to show the main differences between these models and the model that is intended to be published.

Kang, W., Xi, N., & Sparks, A. (2000, April). Formation control of autonomous agents in 3D workspace. In Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No. 00CH37065) (Vol. 2, pp. 1755-1760). IEEE.

Lima, D. A., & Oliveira, G. M. B. (2017, October). Formal analysis in a cellular automata ant model using swarm intelligence in robotics foraging task. In 2017 IEEE International Conference on Systems, Man, and Cybernetics (SMC) (pp. 1793-1798). IEEE.

Rosales, C., Leica, P., Sarcinelli-Filho, M., Scaglia, G., & Carelli, R. (2016). 3d formation control of autonomous vehicles based on null-space. Journal of Intelligent & Robotic Systems, 84(1), 453-467.

McCord, Cassandra, et al. "Distributed progressive formation control for multi-agent systems: 2d and 3d deployment of uavs in ros/gazebo with rotors." 2019 European Conference on Mobile Robots (ECMR). IEEE, 2019.

Paul, T., Krogstad, T. R., & Gravdahl, J. T. (2008, June). UAV formation flight using 3D potential field. In 2008 16th Mediterranean Conference on Control and Automation (pp. 1240-1245). IEEE.

Fathian, K., Safaoui, S., Summers, T. H., & Gans, N. R. (2019, May). Robust 3D distributed formation control with collision avoidance and application to multirotor aerial vehicles. In 2019 International Conference on Robotics and Automation (ICRA) (pp. 9209-9215). IEEE.

Aranda, M., López-Nicolás, G., Sagüés, C., & Zavlanos, M. M. (2014, September). Three-dimensional multirobot formation control for target enclosing. In 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems (pp. 357-362). IEEE.

Ramirez, B., Chung, H., Derhamy, H., Eliasson, J., & Barca, J. C. (2016, November). Relative localization with computer vision and uwb range for flying robot formation control. In 2016 14th International Conference on Control, Automation, Robotics and Vision (ICARCV) (pp. 1-6). IEEE.

Vidal, R., Shakernia, O., & Sastry, S. (2003, September). Formation control of nonholonomic mobile robots with omnidirectional visual servoing and motion segmentation. In 2003 IEEE International Conference on Robotics and Automation (Cat. No. 03CH37422) (Vol. 1, pp. 584-589). IEEE.

 

Author Response

Thank you very much for your valuable comments. The response to Reviewer 1 is attached.

Author Response File: Author Response.pdf

Reviewer 2 Report

The article is devoted to autonomous robots (AR), which can calculate any of their close ARs using proximity sensors and self-organize them into various 3D forms. The proposed approaches lead to a given three-dimensional shape without requiring global localization of the AR, except for the leader.

However, I have several questions for the Author. Please, find them below.

1. The introduction lacks an overview of other path planning algorithms for 3D robot formations. (for example, "3D robot formations path planning with fast marching square", "3D formation control of autonomous vehicles based on null-space", "Implicit coordination for 3D underwater collective behaviors in a fish-inspired robot swarm"

2. Does the algorithm consider various obstacles (not other ARs) when generating a 3D shape?

3. Are you familiar with the recent work: "CARE: A collision-aware mobile robot navigation in grid environment using improved breadth first search"? The authors of this article use an improved breadth-first search algorithm for navigating mobile robots. Maybe it can improve your work.

4. Why is dt (sampling interval) = 1 sec selected (Table 1), isn't it too big? Is this value chosen to speed up the simulation in Matlab or is this how the system will work in reality?

5. Please explain why 70 is chosen in formula 9 (~Line 389)? is it a universal constant for any parameters?

 

Minor issues:

 

1. Line 44-45. Two sentences begin with the phrase: References [9, 10].. it is better to combine these sentences or rephrase them.

2. Line 26, 49, 184. References [2-4],[12-15], and [21-25] are not described in the text, please briefly describe these studies.

3. Line 320. Perhaps the character is not supported.

4. It would be nice to add numerical estimates of the study's achieved results to the conclusion section.

However, I believe that the peer-reviewed manuscript could be a great contribution to MDPI after relatively minor revisions.

 

Author Response

Thank you very much for your valuable comments. The response to Reviewer 2 is attached.

Author Response File: Author Response.pdf