Introduction

1. Related Work

2. Methodology

3. The Simulator

Figure 1. Screenshot of the simulator described in [13].

Figure 1. Screenshot of the simulator described in [13].

3.3. Transformation of Belief and Confidence Maps

In Figure 2, it can be observed that the coordinate systems used by the respective platforms can differ significantly. To ensure that future HSI clients can connect to the simulation, the simulation does not transform the raw data of the belief and confidence maps. Each HSI client must transform the map, if necessary, before using it correctly.

In order to use the information from the belief map in the desktop HSI (Java), the belief map needs to be mirrored along the x- and y-axis (see transformation matrix in 1).

$$\begin{array}{c}\hfill \left[\begin{array}{cc}-1& 0\\ 0& -1\end{array}\right]\left[\begin{array}{cc}1& 0\\ 0& -1\end{array}\right]\left[\begin{array}{c}x\\ y\end{array}\right]=\left[\begin{array}{c}-x^{\prime }\\ y^{\prime }\end{array}\right]\end{array}$$

(1)

The confidence map is created slightly differently by the simulation. To use the information from it as well, the 2D array of the confidence map needs to be rotated by 90 degrees and then mirrored along the y-axis by the VR HSI and desktop HSI (see transformation matrix in 2).

$$\begin{array}{c}\hfill \left[\begin{array}{cc}-1& 0\\ 0& -1\end{array}\right]\left[\begin{array}{c}x\\ y\end{array}\right]=\left[\begin{array}{c}-y^{\prime }\\ x^{\prime }\end{array}\right]\end{array}$$

(2)

Figure 2. Overview of the coordinate systems used by each platform.

Figure 2. Overview of the coordinate systems used by each platform.

4. The Swarm Algorithm

Figure 3. A map of a mission area with UAVs, a disaster, and a human operator. All agents (UAVs and the operator) have belief and confidence maps with a lower resolution (40 × 40) compared to the actual images (600 × 600). The operator uses the interface to communicate with the UAVs in its local neighborhood. This is a direct copy from [13].

Figure 3. A map of a mission area with UAVs, a disaster, and a human operator. All agents (UAVs and the operator) have belief and confidence maps with a lower resolution (40 × 40) compared to the actual images (600 × 600). The operator uses the interface to communicate with the UAVs in its local neighborhood. This is a direct copy from [13].

4.1. Exploratory Behavior

To apply the concept of pheromone trails in ant colonies [7] to our UAVs, the algorithm from [13] was slightly modified. In addition to the already implemented random flight behavior of the UAVs, a straight random flight behavior was implemented. The implementation can be found in the flowchart in Figure 4. Based on a predetermined probability (indicated by a red number in Figure 4), it is decided whether the new flight direction will be randomly determined for the next 30 steps. Afterward, the UAV can decide whether to follow its own exploratory behavior patterns or apply linear exploratory behavior again.

Figure 4. Flowchart for the exploratory behavior algorithm.

Figure 4. Flowchart for the exploratory behavior algorithm.

5. Network Interface

5.1. UDP Packets

Once all events within the simulation are completed after a tick, all relevant information for the HSI is retrieved and prepared for network transmission. Table 1 provides an overview of all parameters that are shared over the network.

Table 1. List of all parameters transmitted over UDP by the network interface after each tick.

Table 1. List of all parameters transmitted over UDP by the network interface after each tick.

Parameter	Description
drones	The number of all UAVs in the current simulation
drones_x	An array with all current X positions of the respective UAVs
drones_y	An array with all current Y positions of the respective UAVs
humans	The number of all human operators in the current simulation
humans_x	An array with all current X positions of the human operators
humans_y	An array with all current Y positions of the human operators
health	The average health of all UAVs
known_drones	The number of UAVs detected by the swarm
disaster	An array with the positions of all disasters in the current simulation
internal_map	The internal belief map of the human operator
confidence_map	The internal confidence map of the human operator

To ensure a disturbance-free network connection and reliable data transmission, the UDP protocol was chosen at this point. Unlike the TCP protocol, it does not check whether all data packets have reached the destination. This approach is justifiable in this case because the high frequency of ticks and thus sent data packets can compensate for any missing packets due to the large number of subsequent ticks. This allows for faster and more efficient data transmission and ensures smooth and delay-free simulation. The parameters marked in blue represent internal information. These are pieces of information available to the swarm. The other parameters pertain to the representation of the simulation in the respective HSI, providing external information accessible to the user but not the swarm algorithms. This is necessary as the current implementation is limited to the simulation, and these parameters would be omitted when testing the system in a real-world environment.

6. The VR-HSI

Table 2. Overview of all scripts of the VR-HSI.

6.2. Beam

The script Beam provides the user with the only input method. Through a light beam, the user can mark areas in their environment (Human pointing model [9]). The script then calculates the marked world coordinates in such a way that the selected position corresponds to the trust map. Subsequently, a potential field is drawn in the trust map starting from this marked point. Depending on the selected tool, the marked point exerts either attractive or repulsive forces on the UAVs.

To transmit the input to the simulator, a TCP connection must first be established on port 65432. After the user confirms the input, the trust map is converted to JSON format and sent to the simulation.

Figure 5. Exemplary configuration of network traffic between simulation and VR-HSI.

Figure 5. Exemplary configuration of network traffic between simulation and VR-HSI.

7. Desktop-HSI

7.1. User Interface

The left map (belief map) marks the positions of disasters and emergency forces. Red squares represent disasters, while red crosses symbolize emergency forces. The overview displays the number of disasters, emergency forces, known drones, and the operational capability of the swarm, indicated by an estimated value of the remaining energy for the UAVs.

The trust map on the right side is the only means to make statements about the positions of the UAVs. This is because the interface only has access to internal information available to the swarm. By clicking on the "Mark Area" button, the selection tools are activated. In the settings on the right side, you can choose between a blue or red circle. The blue circle attracts the UAVs from the selected point, while the red circle repels the UAVs from the selected point. Once the correct tool is selected, a point can be marked on the left map using the mouse. By clicking the "Apply Area" button, the input is sent and executed.

Figure 6. On the left, the user interface of the VR-HSI, and on the right, the user interface of the Desktop-HSI.

Figure 6. On the left, the user interface of the VR-HSI, and on the right, the user interface of the Desktop-HSI.

8. Experiment

Figure 7. Questionnaire used for the pre-survey. Adapted from [14].

Figure 7. Questionnaire used for the pre-survey. Adapted from [14].

Figure 8. Questionnaire used in the survey. Adapted and modified from [14].

Figure 8. Questionnaire used in the survey. Adapted and modified from [14].

9. Data Preparation

10. T-Test

Figure 9. Questionnaire used in the survey. Adapted and modified from [14].

Figure 9. Questionnaire used in the survey. Adapted and modified from [14].

11. Data Preparation

12. T-Test

Table 3. Results of the t-test. Significance levels are marked with (*) symbol. Data set from Figure 9.

13. Mann-Whitney U Test

Table 4. Results of the Mann-Whitney U test. Significance levels are marked with (*) symbol. Data set from Figure 9.

14. Further Analytical Methods

15. Investigation of Exploratory Behavior

Table 5. Results of the t-test for occurrence probabilities of linear exploratory behavior. The differences in occurrence probability are always compared to 0% occurrence probability. Mean and standard deviation are in seconds.

16. Sample Validation

16.1. Computer Experience

As can be seen from Table 6 and the mean of 0.98, it can be assumed that all participants are well versed in computer usage. This assumption is further supported by the low standard deviation of 0.07, indicating that there are minimal differences between the participants’ levels of knowledge.

Table 6. Evaluation of the question: How often do you use computers?

Table 6. Evaluation of the question: How often do you use computers?

Figure	Mean	Std. Deviation
Figure 10	0.98	0.07

Figure 10. How often do you use computers? 0-Never, 1-Every day.

Figure 10. How often do you use computers? 0-Never, 1-Every day.

16.2. Experience with Robots and Multi-Robot Systems

The query of participants’ frequency of interaction with robots and robot swarms aims to determine whether the participants are already familiar with possible behaviors of these systems. The results are shown in Table 7 and indicate that, on average, the respondents interact occasionally with robots and rarely with robot swarms. However, the high standard deviation of 0.29 and 0.35 respectively suggests that a few participants are more engaged with robots and robot swarms and may respond better to the behavior of these systems (see Chapter 23.1 for more details).

Table 7. Evaluation of the questions: How often do you interact with robots? and How often do you interact with robot swarms or multi-robot systems?

Table 7. Evaluation of the questions: How often do you interact with robots? and How often do you interact with robot swarms or multi-robot systems?

Figure	Mean Value	Standard Deviation
Figure 11	0.59	0.29
Figure 12	0.34	0.35

Figure 11. How often do you interact with robots? 0-Never, 1-Every day.

Figure 11. How often do you interact with robots? 0-Never, 1-Every day.

Figure 12. How often do you interact with robot swarms or multi-robot systems? 0-Never, 1-Every day.

Figure 12. How often do you interact with robot swarms or multi-robot systems? 0-Never, 1-Every day.

16.3. Control of Robots and Robot Swarms

The results in Table 8 indicate that the participants feel significantly more confident in controlling individual robots with a mean value of 0.66 compared to controlling robot swarms, where the mean value is only 0.39. Although some experience in controlling robot swarms exists, it is less pronounced. The two high standard deviations suggest a greater variation in the participants’ level of experience within the sample.

Table 8. Evaluation of the questions: How would you rate your experience in controlling robots? and How would you rate your experience in controlling robot swarms?

Table 8. Evaluation of the questions: How would you rate your experience in controlling robots? and How would you rate your experience in controlling robot swarms?

Figure	Mean Value	Standard Deviation
Figure 13	0.66	0.30
Figure 14	0.39	0.34

Figure 13. How would you rate your experience in controlling robots? 0-No experience, 1-Expert.

Figure 13. How would you rate your experience in controlling robots? 0-No experience, 1-Expert.

Figure 14. How would you rate your experience in controlling robot swarms? 0-No experience, 1-Expert.

Figure 14. How would you rate your experience in controlling robot swarms? 0-No experience, 1-Expert.

16.4. Handling VR Technologies and Video Games

The goal is to investigate whether concepts of VR technology, such as a simple human pointing model2 [9], can be used without further training or if participants already have experience from other areas such as video games or VR applications. Whether there is a connection between VR and video game experience in relation to the experiment will be summarized in Chapter 23.1. The level of experience in video game usage is twice as high as in the use of VR applications (see Table 9). Therefore, it can be assumed that the participants already have a good understanding of interactive multimedia techniques. However, the individual participants’ experience values are widely spread due to the two high standard deviations.

Table 9. Evaluation of the questions: How would you rate your experience with video games? and How would you rate your experience with VR/AR technologies in relation to video games or other applications?

Table 9. Evaluation of the questions: How would you rate your experience with video games? and How would you rate your experience with VR/AR technologies in relation to video games or other applications?

Figure	Mean Value	Standard Deviation
Figure 15	0.60	0.29
Figure 16	0.30	0.35

Figure 15. How would you rate your experience with video games? 0-No experience, 1-Expert.

Figure 15. How would you rate your experience with video games? 0-No experience, 1-Expert.

Figure 16. How would you rate your experience with VR/AR technologies in relation to video games or other applications? 0-No experience, 10-Expert.

Figure 16. How would you rate your experience with VR/AR technologies in relation to video games or other applications? 0-No experience, 10-Expert.

17. Effectiveness

Table 10. Overview of successful and unsuccessful missions.

Table 11. Average time to complete a mission.

18. Accuracy

Table 12. Evaluation of Figure 18

Figure 17. Which interface was more accurate in processing your inputs?

Figure 17. Which interface was more accurate in processing your inputs?

Figure 18. How do you rate the quality of communication regarding the instructions you provided to the swarm? 0-My instructions were not executed, 1-My instructions were executed precisely.

Figure 18. How do you rate the quality of communication regarding the instructions you provided to the swarm? 0-My instructions were not executed, 1-My instructions were executed precisely.

19. Response Time

Figure 19. Were your commands executed promptly or within a relevant timeframe? 0-No, 1-Executed immediately and were relevant.

Figure 19. Were your commands executed promptly or within a relevant timeframe? 0-No, 1-Executed immediately and were relevant.

Table 13. Statistics for Figure 19.

20. Usability

20.1. Information about the Swarm

The participants were asked to assess the current objective of the swarm. This could either be exploratory behavior or monitoring a disaster. In contrast to the VR-HSI, the Desktop-HSI could only display information that it had received from the UAVs. However, if no UAV was in communication range with the HSI, the participants could not make any statements about the position of the swarm or its current behavior. This is also reflected in the low mean value of 0.49 for the Desktop-HSI (see Table 14 and Figure 20), which also shows a downward trend. However, one participant was very satisfied with the representation of the Desktop-HSI. This can be attributed to the fact that all disasters were within their communication range and the UAVs were more frequently present there.

In comparison, the VR-HSI is relatively close to the mean value of 0.61. Even without UAVs in the communication range, the participants were able to locate the UAVs on the horizon and determine their flight direction. Additionally, by observing the smoke plumes of the disasters, they could infer the current objective of the UAVs.

Figure 20. Were you always aware of the current objective of the swarm? 0-I didn’t know what was going on, 1-I always knew and could even give instructions to achieve the objective.

Figure 20. Were you always aware of the current objective of the swarm? 0-I didn’t know what was going on, 1-I always knew and could even give instructions to achieve the objective.

Table 14. Statistics for Figure 20.

Table 14. Statistics for Figure 20.

HSI	Mean Value	Standard Deviation
Desktop	0.49	0.27
VR	0.61	0.16

20.2. Information about the Mission Objective

The objective of the mission was to locate four disasters in the mission area. The UAVs marked the respective areas on their belief maps and communicated this information to the HSI. On the Desktop-HSI, potential disasters were marked with a red square on the mission area map. However, the participants could only rely on the information provided by the UAVs. The positions of unknown disasters could only be estimated by the participants on the map, which was perceived as very inaccurate. Most participants were able to orient themselves based on the marked positions in the mission area, but some had difficulty interpreting the beacons on the map and assigning them an accurate position in the mission area.

In contrast, the VR-HSI directly placed beacons at the corresponding positions in the participants’ field of view. This facilitated the spatial interpretation for the participants and helped them better understand the locations of the disasters and their own position. The more precise input (see Chapter 18) to communicate self-located disasters to the swarm was also preferred by the participants.

Overall, the VR-HSI performed significantly better with a mean value of 0.74 in the evaluation compared to the Desktop-HSI (see Table 18), indicating that the spatial representation and more precise input were advantageous for the participants.

Table 15. Statistics for Figure 21.

Table 15. Statistics for Figure 21.

HSI	Mean Value	Standard Deviation
Desktop	0.56	0.20
VR	0.74	0.16

Figure 21. How do you rate the quality of communication regarding the information provided by the swarm? 0-I didn’t know what to do with the information, 10-I could directly see where a disaster was and where the swarm was approximately located.

Figure 21. How do you rate the quality of communication regarding the information provided by the swarm? 0-I didn’t know what to do with the information, 10-I could directly see where a disaster was and where the swarm was approximately located.

20.3. User-friendliness of the Interface

The user-friendliness of the interface refers to how well the various UI elements are understood and structured. Table 16 shows that both interfaces were rated as highly user-friendly by the participants and had only minimal differences.

Although the majority of participants indicated that they were not particularly familiar with VR interfaces (more on this in the Discussion chapter), the VR-HSI performed better in terms of user-friendliness than the Desktop-HSI. However, it is important to note that six out of eight participants required further familiarization with the concepts of VR GUIs3 The Desktop GUI was immediately understood by all participants, which can be attributed to their daily use of computers (see Figure 10). After trying both interfaces, the majority of participants found the VR-HSI easier to use (see Figure 23).

Table 16. Statistics for Figure 22

Table 16. Statistics for Figure 22

HSI	Mean Value	Standard Deviation
Desktop	0.74	0.13
VR	0.81	0.12

Figure 22. How do you rate the clarity and user-friendliness of the communication and collaboration functions? 0-Not clear, 1-Very clear.

Figure 22. How do you rate the clarity and user-friendliness of the communication and collaboration functions? 0-Not clear, 1-Very clear.

Figure 23. Which interface was easier to use?

Figure 23. Which interface was easier to use?

21. Immersion

Table 17. Statistics for Figure 24.

Table 18. Statistics for Figure 25.

Figure 24. Did you feel like being part of the swarm? 0-I only felt like an observer, 1-I received and transmitted information.

Figure 24. Did you feel like being part of the swarm? 0-I only felt like an observer, 1-I received and transmitted information.

Figure 25. Were you able to get a sense of the situation with the given information? 0-I couldn’t learn anything about my surroundings, 1-I knew everything about my surroundings after the mission.

Figure 25. Were you able to get a sense of the situation with the given information? 0-I couldn’t learn anything about my surroundings, 1-I knew everything about my surroundings after the mission.

Figure 26. (a) Which interface gave you a better sense of the robots? and (b) Which interface allowed you to better understand the mission environment?

Figure 26. (a) Which interface gave you a better sense of the robots? and (b) Which interface allowed you to better understand the mission environment?

22. Load

Figure 27. Which interface would you prefer for frequent use?

Figure 27. Which interface would you prefer for frequent use?

23. Interpretation

Figure 28. Left: The VR-HSI interprets the values from the trust map and places a beacon next to a disaster. Right: The Desktop-HSI marks significant values on the trust map with a red square.

Figure 28. Left: The VR-HSI interprets the values from the trust map and places a beacon next to a disaster. Right: The Desktop-HSI marks significant values on the trust map with a red square.

Figure 29. The figure compares for each participant their answers regarding VR prior experience from Figure 16 (blue) with their answers regarding the clarity and user-friendliness of the VR-HSI from Figure 22 (red).

Figure 29. The figure compares for each participant their answers regarding VR prior experience from Figure 16 (blue) with their answers regarding the clarity and user-friendliness of the VR-HSI from Figure 22 (red).

23.1. Influence of Prior Experience on VR and Desktop HSI

To examine whether the different levels of experience of the participants in handling and controlling robotic systems, as well as VR and computer gaming technology, have an influence on VR and Desktop HSI, the participants were divided into two groups. Participants with a rating of $\ge 6$ were classified as experienced, while all others were considered inexperienced. The threshold of $\ge 6$ was chosen because with a threshold of $\ge 5$, there were not enough data available for comparing all criteria due to the small sample size. The responses of both groups were compared in terms of efficiency and accuracy and tested for significance (see Table 19). Although a t-test is not suitable for this sample size (see Chapter 12), it was still used to make the results more comparable for future tests with larger samples.

Table 19. To calculate the significance between the inexperienced and experienced participants, a pairwise t-test was conducted. The data from VR-HSI were used for the metrics. Below is an overview of the results of the significance tests.

Table 19. To calculate the significance between the inexperienced and experienced participants, a pairwise t-test was conducted. The data from VR-HSI were used for the metrics. Below is an overview of the results of the significance tests.

Metrics	Criteria	Inexperienced		Experienced		t	p
		Mean	Std. Dev.	Mean	Std. Dev.
Effectiveness	Xp. with robots	4.15	3.61	3.07	1.70	0.39	0.732
	Xp. robot swarms	4.94	4.72	2.75	0.25	0.69	0.614
	Control of robots	5.10	4.50	1.70	0.14	1.04	0.489
	Control of robot swarms	4.94	4.72	2.75	0.25	0.69	0.614
	Xp. computer games	5.07	3.18	1.94	0.43	1.51	0.271*
	Xp. with VR	5.04	4.58	2.26	0.93	0.71	0.606
Accuracy	Xp. with robots	0.70	0.10	0.67	0.32	0.18	0.874
	Xp. robot swarms	0.70	0.14	0.70	0		1
	Control of robots	0.70	0.14	0.80	0	-1	0.500
	Control of robot swarms	0.70	0.14	0.70	0		1
	Xp. computer games	0.57	0.25	0.83	0.06	-2.22	0. 157*
	Xp. with VR	0.70	0.14	0.75	0.07	-0.33	0.795

Although no result in Table 19 falls below the significance level of p=0.05, a potential significance could be suspected for the criteria marked with (*). By increasing the sample size, the p-value of these criteria could further decrease, potentially revealing an impact of gaming experience on accuracy and effectiveness. Figure 30 reinforces this suspicion. On average, participants with more gaming experience achieved better completion times for the mission and attributed higher accuracy to the VR-HSI.

For the Desktop-HSI, no significant influences of prior experience can be observed (see Table 20).

Figure 30. Comparison of effectiveness and accuracy of the VR-HSI among participants with different levels of experience in computer games. Each point on the figures represents a participant, where the blue points represent inexperienced participants and the red points represent experienced participants.

Figure 30. Comparison of effectiveness and accuracy of the VR-HSI among participants with different levels of experience in computer games. Each point on the figures represents a participant, where the blue points represent inexperienced participants and the red points represent experienced participants.

Table 20. To calculate the significance between inexperienced and experienced participants, a pairwise t-test was conducted using data from the Desktop-HSI. Below is an overview of the results of the significance tests.

Table 20. To calculate the significance between inexperienced and experienced participants, a pairwise t-test was conducted using data from the Desktop-HSI. Below is an overview of the results of the significance tests.

Metrics	Criteria	Inexperienced		Experienced		t	p
		Mean	Std. Dev.	Mean	Std. Dev.
Effectiveness	Xp. with robots	3.58	0.53	4.28	3.17	-0.33	0.774
	Xp. with robot swarms	3.35	0.49	3.48	0.78	-0.61	0.651
	Control of robots	3.21	0.30	5.79	2.96	-1.37	0.401
	Control of robot swarms	3.35	0.49	3.48	0.78	-0.61	0.651
	Xp. computer games	2.78	0.79	4.88	2.62	-1.07	0.397
	Xp. with VR	5.44	3.45	3.31	0.55	0.75	0.589
Accuracy	Xp. with robots	0.70	0.10	0.63	0.29	0.38	0.742
	Xp. with robot swarms	0.65	0.07	0.60	0.28	0.33	0.795
	Control of robots	0.75	0.21	0.75	0.07	0	1
	Control of robot swarms	0.65	0.07	0.60	0.28	0.33	0.795
	Xp. computer games	0.77	0.15	0.60	0.26	0 .76	0.525
	Xp. with VR	0.70	0.14	0.55	0.21	0.60	0.656

23.2. Influence of User Experience on Performance

The metrics can be divided into two categories: Performance and User Experience. Accuracy, Effectiveness, and Reaction Time fall into the Performance category, while User Friendliness, Immersion, and Workload belong to User Experience. The first category refers to the actual technical value that the HSI provides in dealing with the UAV swarm. The metrics used in this category measure how well an HSI performs in operation and determine whether the HSI can be used for a task or not. On the other hand, user experience depends on the user themselves and is often a subjective assessment. However, positive trends in user friendliness can also have a positive impact on performance, as seen in Table 21. In particular, experiences in the Immersion and Workload metrics show an overall more positive rating of the performance metrics by the participants.

Table 21. The summary of the results shows whether participants who had a good user experience also exhibited higher performance in the respective metrics. Values marked with an asterisk (*) indicate that a positive user experience had a positive impact on performance (M stands for mean). Participants who rated above 5 in the respective metric indicated a positive user experience.

Table 21. The summary of the results shows whether participants who had a good user experience also exhibited higher performance in the respective metrics. Values marked with an asterisk (*) indicate that a positive user experience had a positive impact on performance (M stands for mean). Participants who rated above 5 in the respective metric indicated a positive user experience.

User Experience	Performance	Negative		Positive
		M-VR	M-Desktop	M-VR	M-Desktop
	Effectiveness	2.69	4.23	3.362	3.248
	Accuracy	0.67	0.69	0.72*	0.68
User Friendliness	Reaction Time	0.74	0.63	0.72	0.66*
	Effectiveness	2.90	3.63	3.43	3.8
	Accuracy	0.63	0.6	0.72*	0.7*
Immersion	Reaction Time	0.7	0.575	0.715*	0.7*
	Effectiveness	3.46	4.21	3.855	2.66
	Accuracy	0.55	0.52	0.75*	0.85*
Workload	Reaction Time	0.6	0.75	0.72*	0.8*

24. Limitations

Conclusion

Abbreviations

References

1. Barca, J.C.; Sekercioglu, Y.A. Swarm robotics reviewed. Robotica 2013, 31, 345–359. [Google Scholar] [CrossRef]
2. Castelló Ferrer, E. A Wearable General-Purpose Solution for Human-Swarm Interaction. In Proceedings of the Future Technologies Conference (FTC) 2018; Arai, K., Bhatia, R., Kapoor, S., Eds.; Springer International Publishing: Cham, 2019; pp. 1059–1076. [Google Scholar]
3. Hexmoor, H.; McLaughlan, B.; Baker, M. Swarm Control in Unmanned Aerial Vehicles. IC-AI 2005, 911–917. [Google Scholar]
4. Mills, K.L. A brief survey of self-organization in wireless sensor networks. Wireless Communications and Mobile Computing 2007, 7, 823–834. [Google Scholar] [CrossRef]
5. Wong, T.; Tsuchiya, T.; Kikuno, T. A self-organizing technique for sensor placement in wireless micro-sensor networks. In Proceedings of the18th International Conference on Advanced Information Networking and Applications; 2004; Volume 1, pp. 78–83. [Google Scholar] [CrossRef]
6. Berman, S.; Lindsey, Q.; Sakar, M.S.; Kumar, V.; Pratt, S.C. Experimental Study and Modeling of Group Retrieval in Ants as an Approach to Collective Transport in Swarm Robotic Systems. IEEE 2011, 99, 1470–1481. [Google Scholar] [CrossRef]
7. Arvin, F.; Krajník, T.; Turgut, A.E.; Yue, S. COSF: Artificial pheromone system for robotic swarms research. In Proceedings of the 2015 IEEE/RSJ international conference on intelligent robots and systems (IROS), 2015; IEEE; pp. 407–412. [Google Scholar]
8. Xiaoning, Z. Analysis of military application of UAV swarm technology. In Proceedings of the 2020 3rd International Conference on Unmanned Systems (ICUS), 2020; pp. 1200–1204. [Google Scholar] [CrossRef]
9. Argelaguet, F.; Andujar, C. A survey of 3D object selection techniques for virtual environments. Computers & Graphics 2013, 37, 121–136. [Google Scholar] [CrossRef]
10. Chen, M.; Zhang, P.; Wu, Z.; Chen, X. A multichannel human-swarm robot interaction system in augmented reality. Virtual Reality & Intelligent Hardware 2020, 2, 518–533. [Google Scholar] [CrossRef]
11. Suresh, A.; Schwager, M. Brain-Swarm Interface (BSI): Controlling a Swarm of Robots with Brain and Eye Signals from an EEG Headset. 2016. [Google Scholar] [CrossRef]
12. Triantafyllidis, E.; Mcgreavy, C.; Gu, J.; Li, Z. Study of Multimodal Interfaces and the Improvements on Teleoperation. IEEE Access 2020, 8, 78213–78227. [Google Scholar] [CrossRef]
13. Divband Soorati, M.; Clark, J.; Ghofrani, J.; Tarapore, D.; Ramchurn, S.D. Designing a User-Centered Interaction Interface for Human–Swarm Teaming. Drones 2021, 5. [Google Scholar] [CrossRef]
14. Oradiambalam Sachidanandam, S. Effectiveness of Augmented Reality Interfaces for Remote Human Swarm Interaction. PhD thesis, 2021. [Google Scholar] [CrossRef]