**2. Challenges**

The contributions to the Special Issue reveal a good amount of challenges for the operation of MRSs. Some are related to multi-agent mission planning, intervention in complex scenarios with uncertainty, and operation under restricted communications. Robot cooperation often helps deal with these issues through sharing information and coordinating actions. Kurt Geihs [1] reviews the state-of-the-art on teamwork in the context of multi-robot systems and dynamic environments. His paper identifies and analyzes multiple engineering challenges: dynamic coalitions, platform harmonization and configuration, knowledge base, methodology and tools, edge and cloud integration, and human in the loop and other sociotechnical concerns, among other concepts.

Rongye Shi, Peter Steenkiste, and Manuela M. Veloso [2] address the challenge of Multi-Agent Path Planning (MAPP), which is a resource allocation problem complicated by the highly dynamic and distributed environments. In contrast to most MAPP approaches, they assume soft collisions instead of hard collisions, which means that the agents can share resources or concur at the same location at the expense of reducing the quality of solutions. They propose the Soft-Collision M\* (SC-M\*) algorithm to solve these constrain satisfaction problems and compare its results with the most common algorithms in terms of path cost, success rate, and run time. Meanwhile, Yang Lyu, Quan Pan, and Jian Lv [3] address the problem of multi-robot collaborative self-localization in the context of target tracking missions. They propose an unscented transformation-based collaborative self-localization algorithm to deal with inter-robot and robot–target correlations during the missions.

Martin Juhás and Bohuslava Juhásová [4] address other relevant challenges for MRSs, which are communications and coordination. Their paper presents a time-synchronization solution for operations performed by a heterogeneous set of robotic manipulators grouped into a production cell. They develop a master–slave architecture without an external control element, whose communications are implemented via TCP/IP sockets. Similarly, Facundo Benavides, Caroline Ponzoni Carvalho Chanel, Pablo Monzón, and Eduardo Grampín [5] consider the multi-robot exploration problem under restricted communications. They propose a novel auto-adaptive multi-objective strategy to support the selection of tasks regarding both exploration performance and connectivity level. Compared with other algorithms, it shows effectiveness and flexibility to tackle the multi-robot exploration problem, decreasing the effects of disconnection periods without noticeable degradation of the exploration time.

Cooperation can be carried out in homogeneous fleets but also between robots with different morphology. Pablo R. Palafox, Mario Garzón, João Valente, Juan Jesús Roldán, and Antonio Barrientos [6] analyze air–ground robot cooperation in their paper. The combination of aerial and ground robots is useful in search and rescue tasks, given that aerial robots can provide valuable insight to support the navigation of ground robots in complex scenarios affected by disasters. The article proposes a state machine with algorithms that allow an aerial robot to take off, track, and land on a mobile ground platform.

Abhijeet Ravanka, Ankit A. Ravankar, Yohei Hoshino, and Yukinori Kobayashi [7] focus on another relevant challenge in multi-robot missions: uncertainty. They find information sharing as a powerful tool to deal with uncertainty in mission planning. In this way, when a robot finds a new obstacle or blocked path, it can share this information with the rest of the fleet, allowing other robots to plan better paths. The paper proposes a novel method for information sharing that works when robots have different sensors, there is positional uncertainty, and obstacles are dynamic. Aliakbar Akbari, Mohammed Diab, and Jan Rosell [8] also focus on uncertainty but in the context of mobile manipulation. In these applications, humans can collaborate with robots to execute complex actions, sharing their knowledge about the task and scenario. They propose a contingent-based task and motion planning method that generates trees of feasible plans considering robot uncertainty and human–robot interactions. This algorithm is validated in grasp tasks with occluding objects.

#### **3. Trends**

Currently, one of the most relevant multi-robot systems is swarms. David Garzón Ramos and Mauro Birattari [9] present an automatic method to design robot swarms. This method can generate control software by assembling preexisting software modules via optimization. They validate these developments with a swarm of e-pucks, which can use color-based information for handling events, communicating, and navigating. James Wilson, Jon Timmis, and Andy Tyrrell [10] address another relevant aspect of swarms: hormone systems for collective behaviors. They use a collection of virtual hormones to control the selection of behaviors that produce an effective foraging swarm.

Immersive technologies such as virtual, augmented, and mixed realities are usually proposed to improve operator workload, situational awareness, and performance. Ashish Kumar, Sugjoon Yoon, and V.R. Sanal Kumar [11] present a mixed reality simulation for Unmanned Aerial Vehicles in high-endurance missions of Earth exploration. This environment combines real and virtual quadcopters to monitor missions and find paths, among other things. Luis Pérez, Silvia Rodríguez-Jiménez, Nuria Rodríguez, Rubén Usamentiaga, and Daniel F. García [12] propose the creation of digital twins of manufacturing processes in Industry 4.0. In this way, a virtual reality testbed can be used to design, implement, and monitor the process in real-time before its physical development.

Finally, machine learning and especially reinforcement learning are increasingly popular for MRSs, especially for their operation in complex unstructured scenarios. Wenzhou Chen, Shizheng Zhou, Zaisheng Pan, Huixian Zheng, and Yong Liu [13] apply deep reinforcement learning for the collaborative formation and navigation of a robot fleet.

#### **4. Applications**

Robots are often applied in emergency scenarios because they can obtain information and even intervene, preventing dangers for human teams. Multi-robot systems are being introduced in these missions, such as outdoor and indoor fires. Juan Jesús Roldán-Gómez, Eduardo González-Gironda, and Antonio Barrientos [14] propose the use of drone swarms for the prevention, surveillance, and extinguishing of forest fires. This system consists of quadcopters that individually can visit waypoints and use payloads but, collectively, can perform complex tasks. The authors propose the use of immersive interfaces to allow operators to control multiple drones simultaneously.

However, emergency scenarios are not the only ones in which these systems can add value. Ángel Montes-Romero, Arturo Torres-González, Jesús Capitán, Maurizio Montagnuolo, Sabino Metta, Fulvio Negro, Alberto Messina, and Aníbal Ollero [15] propose a set of director tools for autonomous media production with a team of drones. They focus on a language for cinematography mission description and a procedure to translate missions into plans, so a media director that is not necessarily familiar with robots can manage the system.

Finally, multi-robot systems have also reached social robotics, although the need was not as clear as in other fields. Toshiaki Nishio, Yuichiro Yoshikawa, Kohei Ogawa, and Hiroshi Ishiguro [16] study multi-party conversations with two human-like robots. They focus on conveying information to the viewers through a natural conversation between the robots. Takamasa Iio, Yuichiro Yoshikawa, Mariko Chiba, Taichi Asami, Yoshinori Isoda, and Hiroshi Ishiguro [17] try a question-answer-response dialogue model with two humanoid robots to involve elderly users in the conversation. The results suggest that the presence of two robots might likely encourage elderly people to sustain longer talks.

**Author Contributions:** Conceptualization, J.J.R.-G. and A.B.; methodology, J.J.R.-G. and A.B.; writing—original draft preparation, J.J.R.-G.; writing—review and editing, J.J.R.-G.; visualization, J.J.R.-G.; supervision, A.B.; project administration, J.J.R.-G. and A.B.; funding acquisition, J.J.R.-G. and A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** Work produced with the support of a 2020 Leonardo Grant for Researchers and Cultural Creators, BBVA Foundation. The Foundation takes no responsibility for the opinions, statements, and contents of this project, which are entirely the responsibility of its authors.

**Acknowledgments:** This Special Issue has been possible thanks to the hard work of authors, reviewers, and editors. We would like to thank and congratulate all the authors for their valuable contributions to our Special Issue. In addition, we would like to express our gratitude to Daria Shi, Managing Editor of Applied Sciences (MDPI), for her outstanding work during these years. Juan Jesús Roldán-Gómez wants to dedicate this book to the memory of their grandmother Gloria Montoya Guerrero, who would have been very proud to hear about it on one of their Sunday evening calls.

**Conflicts of Interest:** The authors declare no conflict of interest.
