Swarm Robotics: Simulators, Platforms and Applications Review

Round 1

Reviewer 1 Report

Add dates to Table 2 to show when the robots were introduced.

Author Response

Point 1: Add dates to Table 2 to show when the robots were introduced.

Response 1: Introduction dates were added on Table 2

Principal changes were marked with yellow in the attachment

Reviewer 2 Report

This manuscript reviews the recent progress on swarm robots in terms of software, hardware, simulators, applications and behaviors. It presents a reference  to explore and research swarm robotic area, and provides options for transitioning from simulations to real swarm robotic implementations. There are some minor issues that need to be addressed before consideration of acceptance.

• Virtual simulation provides an easier, faster and safer way to perform experiments than using real robots. Moreover, different multi-robot simulators have different development objectives to get closer to reality in different ways. However, since software is impossible to simulate all aspects of reality, what are the typical differences about swarm robots betweensimulation and real-time experiments? It is recommended to add clarifications in the manuscript.  
• Most of the current research on swarm robots is limited to simulations and modeling. Theexperiments of real-life swarm robotics are also very In Section 6, the authors mentioned that “However, applying SR in real-life environments is currently limited by existing technology”. What are technical problems needed to be overcome to break through this limitation? Please give a few examples to briefly illustrate.

Author Response

Point 1: Virtual simulation provides an easier, faster and safer way to perform experiments than using real robots. Moreover, different multi-robot simulators have different development objectives to get closer to reality in different ways. However, since software is impossible to simulate all aspects of reality, what are the typical differences about swarm robots between simulation and real-time experiments? It is recommended to add clarifications in the manuscript.  

Response 1: In section 3.2 the following details were added:

"Most of the time, it is not possible to simulate all aspects of reality. For instance, some simulated sensors and actuators are free of noise and interference. Besides, simulators usually do not take into account all environment interactions like friction, luminosity, temperature, wind or dust. Robot internal interactions are also challenging to be considered on simulators. This may include battery life and energy level, components degradation and failure, sensors and actuators limitations due to static friction, backslash, hysteresis or saturation.

(...)

However, real world experiments are time-consuming and error-prone tasks. Gupta et al. [30] mention that currently most real swarm robotics face security issues. This is mainly because there have not been many studies about security protocols adapted to swarm robotics, therefore communications can be intercepted or disturbed by an attacker.

Another issue is communication within the swarm and between the swarm and the environment. For the swarm to be fully autonomous, it should provide its own means of communication, however, the most used technology (ad-hoc WLAN networks) provides communication with a limited range and it is susceptible to individual robot failures. An infrastructure-based network can provide more stable communication but it requires the installation of hardware, like base stations. This issue causes many industrial projects still rely on centralized communication [31].

In addition, Dorigo et al. [2] comment that options related to materials, biohybrid solutions, and new ways of storing and transmitting energy could help address some of the current issues related to the hardware of real robot swarms."

And in section 6 the following summary:

"As been shown, swarm robotics have a big potential to solve (or improve) a large number of problems and situations. However, applying SR in real-life environments is currently limited by existing technology: environment interaction, noise, sensors, actuators, infrastructure-based network, security protocols, energy, components degradation, and failures as was detailed in section 3.2.

Nonetheless, real swarm robotics implementations are important and necessary to test and suggest real solution to real problems. Simulations and virtual reality represent a previous step on a way to the reality."

____________________________________________________________________________________
Point 2: Most of the current research on swarm robots is limited to simulations and modeling. The experiments of real-life swarm robotics are also very In Section 6, the authors mentioned that “However, applying SR in real-life environments is currently limited by existing technology”. What are technical problems needed to be overcome to break through this limitation? Please give a few examples to briefly illustrate.

Response 2: Included in Response 1. Details added in section 3.2 also mention the technical problems and limitations. Besides, at the end of section 5 we added 

"It is also important to mention that most of the time the swarm size depends on the number of robots that companies or research agencies have in stock for arbitrary reasons [31] and robots may not be intended for all swarm applications." 

and in section 6 

"There is a wide variety of real-life swarm robotic platforms. The spectrum of possibilities range from inexpensive and limited robot agents to full of resource commercial robots. However, there is no rule for defining the right swarm size, most of them depend on the creators/developer's possibilities and sponsors."

____________________________________________________________________________________

Grammatical and typo errors were corrected. Principal changes were marked with yellow in the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

In this manuscript, the authors perform a review of the Swarm Robotics (SR) area in terms of software and hardware platforms by giving an overview and focusing on the open challenges.

The paper is a good background review and it results very interesting to follow. The authors fulfilled all main points with clarity and precision. Indeed, the review includes some theory and background analysis, simulated vs. real setup issues, presents a comparison between existing platforms and simulators and gives a review of the main SR applications.

In my opinion, this manuscript is well organized and good considerations were carried out in addressing all aspects. The sections are well organized and all Tables summarize the concepts with an appropriate level of understanding.

I believe this will be a very useful review of the present swarm robotics applications and challenges.

Nevertheless, some minor aspects should be addressed before acceptance:

- line 2 of the abstract: remove the word “behaviors”. It refers to “other applications section” but here is out of context by resulting insignificant and misleading.

- In line 119 the authors declare: “In this subsection, some of the most popular swarm robotics software will be covered” but the presented software examples are not only swarm robotics simulators. The sentence is misleading. Please, remove or modify.

- From the previous concern, the assertion that ODE is a swarm simulator is not true. It could be useful to show works where multi-agent simulations were made (e.g., https://doi.org/10.1117/12.887388, 10.1109/IJCNN.2012.6252664, https://doi.org/10.5772/60545). For example, in these articles, a 3D dynamic simulator, implemented by using ODE, is presented. It is also used to simulate some cooperation scenarios for inducing collaboration in foraging tasks.

- line 156: What “,Mondada’s,” means? Maybe Mondada’s E-puck? Correct the expression.

- Table 2. It could be useful to introduce also for the robotic platforms a column with the Last Hw/Sw maintenance (similar to Last Update for Table 1).

- In lines 295-297 the authors declare: “In this section, we will show some of the latest works made with SR. We will divide our search in two categories (i) those that were made mainly in simulated environments and (ii) those that employed a real physic swarm.”. But the Section does not include any categorization. Please consider the possibility to introduce a Table also for this concern.

- An extensive editing of the English language and style is required. Please carefully proof-read spell check to eliminate typos (line 165: por instead of for, line 297: physic instead of physical, line 299: scenario instead of scenarios, etc.), grammatical errors and duplication of punctuation marks. In the meanwhile, check for missing or redundant blank spaces (e.g. lines 120, 194).

Author Response

Point 1: line 2 of the abstract: remove the word “behaviors”. It refers to “other applications section” but here is out of context by resulting insignificant and misleading.

Response 1: Corrected

____________________________________________________________________________________
Point 2: In line 119 the authors declare: “In this subsection, some of the most popular swarm robotics software will be covered” but the presented software examples are not only swarm robotics simulators. The sentence is misleading. Please, remove or modify.

Response 2: Modified as follows:

"There are many kinds of platforms, software, and algorithm to simulate robots based on different objectives and types of robots. For example, there are 2D and 3D simulators with specific algorithms to design, analyze and compare swarms. In this subsection, some of the most popular software and platform that could be used to simulate swarm robotics will be covered."

____________________________________________________________________________________
Point 3: From the previous concern, the assertion that ODE is a swarm simulator is not true. It could be useful to show works where multi-agent simulations were made (e.g., https://doi.org/10.1117/12.887388, 10.1109/IJCNN.2012.6252664, https://doi.org/10.5772/60545). For example, in these articles, a 3D dynamic simulator, implemented by using ODE, is presented. It is also used to simulate some cooperation scenarios for inducing collaboration in foraging tasks.

Response 3: Actually, in section 6 we wrote: 

"If the main purpose is to test or design a robot for a swarm, taking into account its locomotion, sensors, and physics; then ODE, MSRS, and CoppeliaSim present the most suitable features. These programs allow simulating robot models as similar to real life as possible. However, they were not created to simulate collective swarm behaviors, but some modifications can be done to use them for this purpose."

Now, we also modified the text  in section 3.1 as follows:

"Open Dynamics Engine (ODE) is a dynamics engine capable of multi-agent simulation and could, consequently, be used for robotics simulation."

____________________________________________________________________________________
Point 4: line 156: What “,Mondada’s,” means? Maybe Mondada’s E-puck? Correct the expression.

Response 4: Corrected

____________________________________________________________________________________
Point 5: Table 2. It could be useful to introduce also for the robotic platforms a column with the Last Hw/Sw maintenance (similar to Last Update for Table 1).

Response 5: Introduction dates were added on Table 2

____________________________________________________________________________________
Point 6: In lines 295-297 the authors declare: “In this section, we will show some of the latest works made with SR. We will divide our search in two categories (i) those that were made mainly in simulated environments and (ii) those that employed a real physic swarm.”. But the Section does not include any categorization. Please consider the possibility to introduce a Table also for this concern.

Response 6: We modified the text as follows:

"In this section, we will show some of the latest works made with SR. We will divide our search into two categories (i) those that employed a real life swarm and (ii) those that were made mainly in simulated environments. For each of these categories, if found, we will present the specific platform or software that was used."

and at the end of section 5 we added a table with a swarm robotics applications summary. The applications were categorized in real life swarm and swarm simulation.

____________________________________________________________________________________
Point 7: An extensive editing of the English language and style is required. Please carefully proof-read spell check to eliminate typos (line 165: por instead of for, line 297: physic instead of physical, line 299: scenario instead of scenarios, etc.), grammatical errors and duplication of punctuation marks. In the meanwhile, check for missing or redundant blank spaces (e.g. lines 120, 194).

Response 7: In general, grammatical and typo errors were corrected. 

____________________________________________________________________________________
Principal changes were marked with yellow in the attachment

Author Response File: Author Response.pdf