Abstract
Swarm robotics addresses the design, deployment, and analysis of large groups of robots that collaborate to perform tasks in a decentralized manner. Research in this field has predominantly relied on simulations or small-scale robots with limited sensing, actuation, and computational capabilities. Consequently, despite significant advancements, swarm robotics has yet to see widespread commercial or industrial application. A major barrier to practical deployment is the lack of affordable, modern, and robust platforms suitable for real-world scenarios. Moreover, a narrow definition of what swarm robotics should be has restricted the scope of potential applications. In this paper, we argue that the development of more advanced robotic platforms—incorporating state-of-the-art technologies such as SLAM, computer vision, and reliable communication systems—and the adoption of a broader interpretation of swarm robotics could significantly expand its range of applicability. This would enable robot swarms to tackle a wider variety of real-world tasks and integrate more effectively with existing systems, ultimately paving the way for successful deployment.
1 Introduction
Swarm robotics investigates the design, deployment, and evaluation of large groups of robots that collaborate in a decentralized manner to complete their tasks (Dorigo et al., 2014). Research in swarm robotics has rapidly advanced in recent years. One of the main challenges in the field—designing control software for robot swarms—has been thoroughly investigated, with many successful approaches proposed. In particular, (semi-) automatic design (Francesca and Birattari, 2016; Bredeche et al., 2018; Birattari et al., 2019; Birattari et al., 2020) has proven effective in addressing this challenge, with evolutionary robotics (Floreano et al., 2008; Haasdijk et al., 2014; Trianni and López-Ibáñez, 2015; Divband Soorati and Hamann, 2015; Nolfi, 2021) and automatic modular design (Francesca et al., 2014; Spaey et al., 2020; Hasselmann et al., 2021; Mendiburu et al., 2022; Kuckling et al., 2022; Hasselmann et al., 2023; Salman et al., 2024) receiving extensive attention. Researchers have also proposed hybrid solutions that combine evolution and automatic modular design (Ligot et al., 2020; Cambier and Ferrante, 2022; Hasselmann et al., 2023). A promising development is the recent integration of multi-level modeling into the automatic design of control software (Baumann et al., 2022a; Endo et al., 2023). Beyond (semi-) automatic design, various model-based approaches have been proposed, including probabilistic (Correll and Martinoli, 2007; 2011), space-time continuous (Hamann and Wörn, 2008; Hamann, 2010), property-driven (Brambilla et al., 2014), and curiosity-driven models (Kaiser and Hamann, 2022).
Alongside the development of design methods, a better understanding of the reality gap and its effects (Jakobi et al., 1995; Floreano et al., 2008; Francesca and Birattari, 2016) has enhanced their robustness to the sim-to-real transfer. Notably, research has demonstrated that effects akin to those of the reality gap can be replicated in simulation (Ligot and Birattari, 2020), giving rise to the notion of “pseudo-reality.” A pseudo-reality is a simulation model distinct from the one used during the design process. The underlying idea is that control software capable of successfully transferring from the design-phase model to a pseudo-reality is more likely to cross the reality gap than software that cannot (Ligot and Birattari, 2022).
As noted by several authors (Schranz et al., 2020; Dorigo et al., 2021; Dias et al., 2021; Cheraghi et al., 2022; Kuckling, 2023), significant progress has also been made in other areas of the field. Swarms have been developed using underwater (Zahadat and Schmickl, 2016; Connor et al., 2020; Berlinger et al., 2021) and aerial robots (McGuire et al., 2019; Soria et al., 2020; Pavliv et al., 2021; Schilling et al., 2022); heterogeneous swarms have been explored (Dorigo et al., 2013; Zhu et al., 2024); machine learning techniques—such as deep (Hüttenrauch et al., 2017; Hüttenrauch et al., 2019; Yasuda and Ohkura, 2019) and inverse (Gharbi et al., 2023; Szpirer et al., 2024) reinforcement learning—have been integrated into swarm design; advanced capabilities like SLAM (Kegeleirs et al., 2021; Lajoie and Beltrame, 2023) and computer vision (Verlekar and Joshi, 2017; Kegeleirs et al., 2024a; Kegeleirs et al., 2024b) have been tested in swarm contexts; studies on human-swarm interaction have been conducted (Kolling et al., 2016; Podevijn et al., 2016); and swarms of pico-satellites have been investigated (Pinciroli et al., 2008a; Pinciroli et al., 2008b; Fdhila et al., 2012). Moreover, although this paper focuses on macrorobots—i.e., robots at the centimeter to meter scale—swarm robotics holds considerable promise for applications in nanotechnology (Hauert and Bhatia, 2014; Law et al., 2023), albeit with unique challenges of its own.
However, despite these advances, real-world applications of swarm robotics remain extremely limited. To date, no commercial or industrial deployment of robot swarms has been reported. Swarm robotics appears to be at a critical juncture—understanding the reasons for this lack of application and identifying possible enablers could help steer the field toward practical deployment.
2 Towards applied swarm robotics
2.1 The practical issues
2.1.1 Platform limitations and experimental constraints
A major obstacle to real-world swarm deployment is the lack of modern, affordable, and reliable experimental platforms. Experiments with physical robots remain costly and time-consuming—challenges that scale with swarm size. Researchers also face practical limitations: large spaces are often unavailable or expensive, and acquiring many robots is financially prohibitive. Common platforms used in single- or multi-robot systems are typically too large and costly for swarm use. This has driven the development of swarm-specific platforms that favor low cost and compactness, but at the expense of sensing, actuation, and computational power. The E-Puck (Mondada et al., 2009), Kilobot (Rubenstein et al., 2012), and Crazyflie (Giernacki et al., 2017) are among the most widely used platforms, but they remain limited—plagued by noisy sensors and unreliable actuators. As a result, researchers frequently resort to abstraction (Dorigo et al., 2021), simplifying missions to work around hardware constraints. For instance, two of the most common swarm robotics missions—aggregation and foraging—demonstrate clear limitations.1 Aggregation—robots gathering at a single location—is achievable with minimal capabilities but has limited relevance for real-world applications. Foraging—moving objects from one location to another—has greater potential in domains such as logistics, warehouse operations, or search and rescue. Yet, current platform limitations often force researchers to abstract away essential components, such as object manipulation. In both simulation (Wei et al., 2016; Harwell and Gini, 2018; Song et al., 2020; Jimenez Romero et al., 2024) and physical experiments (Rubenstein et al., 2014; Francesca et al., 2015; Hecker and Moses, 2015; Pitonakova et al., 2018; Talamali et al., 2020), robots typically do not carry real objects, undermining the practical credibility of the task. Moreover, while robot swarms typically operate at high densities in lab experiments, envisioned real-world applications often involve sparse swarms, which would call for different control strategies to perform their intended tasks (Tarapore et al., 2020; Kwa et al., 2023b). This results in a disconnect between the long-term ambitions of swarm robotics—space exploration, search and rescue, ocean cleaning—and the highly abstracted, constrained experiments that currently dominate the field. Compounding this issue, platforms like the E-Puck and Kilobot rely on outdated hardware and software architectures. Although efforts have been made to modernize them, such as the Pi-Puck extension for the E-Puck (Allen et al., 2020) or the ROS-ready operating system DeimOS (Kegeleirs et al., 2025), they still fall short of overcoming fundamental limitations.
Researchers are often left with two choices: build custom robots or rely exclusively on simulation. Custom-built robots are costly and time-consuming to develop, often tailored to specific research needs—limiting reusability and reproducibility. These efforts are frequently undervalued and rarely supported by thorough documentation. Achieving both capability and compactness is challenging: advanced components demand more space or power, resulting in form factors unsuited to large-scale swarms. Moreover, affordability depends on mass production, which custom robots cannot achieve, while miniaturization requires investments only feasible at industrial scale. Newer platforms such as the S-Drone (Oguz et al., 2022), Mercator (Kegeleirs et al., 2022), and Summit XL (Arregi and Secco, 2023) offer improved capabilities, but none have resolved these trade-offs well enough to gain widespread adoption.
2.1.2 Simulation tools and the deployment gap
Resorting exclusively to simulations is not an ideal solution either. Few simulators are well-suited to swarm robotics. ARGoS3 (Pinciroli et al., 2012) and SwarmLab (Soria et al., 2020), while specifically designed for this purpose, lack the extensive documentation and community support of more general-purpose tools like Gazebo (Koenig and Howard, 2004). Gazebo itself, however, is poorly optimized for large-scale swarm simulations, often struggling to handle more than a few dozen robots efficiently. More critically, the reality gap remains a central concern, especially in the evolutionary approach (Jakobi et al., 1995; Hasselmann et al., 2021). Recent applications of reinforcement learning to robot swarms (Hüttenrauch et al., 2017; Hüttenrauch et al., 2019; Yasuda and Ohkura, 2019) appear to face similar challenges. This is suggested by the scarcity of convincing experiments conducted with real robots in the current literature. More generally, even in single-robot contexts, reinforcement learning methods are known to struggle with sim-to-real transfer due to their sensitivity to modeling inaccuracies and environmental variability (Zhao et al., 2020; Salvato et al., 2021). Consequently, studies conducted solely in simulation provide only limited insight into whether the system will function as expected with real robots. Pseudo-reality (Koos et al., 2013; Ligot and Birattari, 2020) might offer a partial mitigation by exposing controllers to model variations, but it cannot fully guarantee real-world reliability.
Recent studies (Kegeleirs et al., 2024c) have shown that even when a design method succeeds on one physical platform, it may fail to transfer to another. This challenge extends beyond the sim-to-real gap and also affects deployment across physical platforms. We refer to this broader issue as the deployment gap: regardless of whether control software is developed in simulation or on a specific robot, its effectiveness is not guaranteed when applied to another platform. Although some methods robust to the reality gap show partial resilience to the deployment gap, further performance degradation still occurs. Consequently, developing control software using overly simplistic robots increases the risk of failure when moving to more capable, field-ready systems.
Resource-sharing infrastructures such as the Robotarium (Wilson et al., 2020) offer partial relief by enabling remote access to real robot swarms. However, these services have notable limitations: users cannot directly interact with the robots or their environment, iterative debugging is more difficult, and demand can restrict timely access. Moreover, the robots—though more modern than Kilobots or E-Pucks—still have limited capabilities, and the system does not easily accommodate complex missions or environmental changes. Scaling up such services would likely require commercial backing, introducing additional costs for users.
2.1.3 Integration challenges: SLAM, vision, and communication
Then, the limitations of current swarm platforms hinder the integration of key robotics technologies such as SLAM, computer vision, and communication. Although its potential was envisioned by early work (Schmickl et al., 2006), only recently has swarm SLAM (Kegeleirs et al., 2021) begun to show practical results (Lajoie and Beltrame, 2023)—and even then, only under constrained, highly structured conditions that are not typical of swarms. Reliable localization remains a significant challenge, especially in the absence of global positioning systems (Quraishi and Martinoli, 2022; Braga et al., 2024). Similarly, while computer vision is ubiquitous in general robotics, it remains underutilized in swarms, largely due to technical limitations. Basic applications have been demonstrated, such as color-based signaling (Nouyan et al., 2009; Chen et al., 2015; Jones et al., 2019; Garzón Ramos and Birattari, 2020) and human-robot interaction based on simple gesture and face recognition (Nagi et al., 2014; Suresh and Martínez, 2019). However, more advanced vision capabilities, like person tracking and re-identification, have only recently been explored in distributed systems (Popovici et al., 2022) and swarms (Kegeleirs et al., 2024a; Kegeleirs et al., 2024b), with modest results so far. Communication also remains underdeveloped (Di Caro et al., 2005; Cianci et al., 2006). Many swarm experiments omit communication altogether, or rely on highly abstracted models such as neighbor detection without actual data exchange. In rare cases, robots share small amounts of numerical data (Ducatelle et al., 2011; Ducatelle et al., 2014; Hasselmann and Birattari, 2020; Garzón Ramos and Birattari, 2020; Talamali et al., 2021; Kuckling et al., 2022). Stigmergy has gained renewed interest (Hunt et al., 2019), but existing implementations often depend on fixed infrastructures that are costly and limited to specific environments (Khaliq et al., 2014; Reina et al., 2021; Na et al., 2021), or materials like wax and alcohol (Russell, 1997; Fujisawa et al., 2014), which pose safety risks due to their flammability and are impractical for most applications. More recently, alternatives based on photo-chromatic pigments have shown potential (Salman et al., 2020; Salman et al., 2024).
Communication between the swarm and external systems—other robots (Dorigo et al., 2013; Kegeleirs et al., 2024b; Zhu et al., 2024) or humans (Nagi et al., 2014; Kolling et al., 2016; Mondada et al., 2016)—is also rare. Again, this is largely due to the lack of suitable communication hardware and protocols. Yet, such capabilities are critical for many envisioned applications.
2.1.4 Regulatory, ethical, and societal barriers
Finally, the deployment of robot swarms remains constrained by ethical and regulatory considerations, particularly regarding their potential ecological and societal impacts (Garzón Ramos and Hauert, 2024; Winfield et al., 2025). This is especially pronounced in the context of aerial drones: although the underlying technology is sufficiently advanced to perform tasks such as aerial surveillance and object detection or recognition, the operation of UAV swarms typically necessitates regulatory waivers and exemptions, which vary significantly across jurisdictions (UK Civil Aviation Authority, 2022; Australian Civil Aviation Safety Authority, 2024; Code of Federal Regulations, 2025). A key factor underlying the reluctance of regulators and operators is the inherent uncertainty surrounding swarm behavior, compounded by a lack of transparency and explainability in their collective decision-making processes (Hussein et al., 2020; Naiseh et al., 2024). Accordingly, it is critical to investigate how swarms are perceived by human users (Carrillo-Zapata et al., 2020) and to develop strategies for fostering (Nam et al., 2019; Lyons et al., 2025) and maintain (Liu et al., 2019) trust, thereby enabling more effective collaboration between humans and robots (Divband Soorati et al., 2022). Public skepticism may also be exacerbated by the increasing use of drones in military contexts, as well as by dystopian portrayals of robots and AI in popular media, both of which may hinder acceptance of swarm technologies in everyday settings.
2.2 The conceptual issues
2.2.1 Rigid interpretations of swarm principles
At a conceptual level, swarm robotics often clings to foundational conventions, treating core principles as fixed rules and defaulting to standard design choices without critical reflection. The canonical definition emphasizes fault tolerance, flexibility, and scalability—emerging from redundancy, self-organization, and locality of sensing and communication (Dorigo et al., 2014). These principles have driven progress but can become limiting when treated as strict requirements. Due to technical and economic constraints, achieving them in practice is often difficult, prompting researchers to simplify experiments just to preserve the “swarm” designation. Moreover, while these features offer clear advantages, they are often seen in industry as impractical compared to centralized, high-performance systems. This rigidity discourages hybrid approaches that might be more viable in real-world settings. A swarm can—and arguably should—leverage centralized components when useful, without losing its distributed character.
2.2.2 Verification and assumptions about swarm properties
Equally important, fault tolerance, flexibility, and scalability are often assumed rather than formally or empirically verified. Formal verification remains a major challenge. Early approaches based on temporal logic (Rouff et al., 2004; Winfield et al., 2005; Dixon et al., 2012; Gjondrekaj et al., 2012) are highly sensitive to the state explosion problem, limiting their scalability. Later methods—such as probabilistic model checking (Konur et al., 2012), statistical model checking (Massink et al., 2013), and property-driven design (Brambilla et al., 2014)—enabled more scalable analyses but often lacked consistent implementation of system models that support practical simulation and testing. More recent techniques come with trade-offs: some require extensive expert knowledge (Coppola et al., 2019), others focus solely on software-level verification (Merlo et al., 2022), and some reintroduce scalability issues (Leofante et al., 2019). Consequently, researchers often fall back on qualitative demonstrations—for example, showing stable performance across different swarm sizes or different environments, or resilience to robot failures. Recent findings indicate that scalability (Kuckling et al., 2024) and possibly other key properties (Hunt et al., 2019) may have practical limitations—even in systems specifically designed to exhibit them. While these properties are definitely an asset of robot swarms, making unexamined assumptions about them without rigorous validation risks misleading future research.
2.2.3 Isolation vs integration with external systems
Rigid thinking in swarm robotics also affects how swarms are composed and interact with other systems. Most studies focus on homogeneous swarms operating in isolation. Even heterogeneous swarms—where different types of robots collaborate (Ducatelle et al., 2011; Dorigo et al., 2013)—are typically treated as self-contained entities (Kwa et al., 2020; Wang et al., 2021; van Diggelen et al., 2024), broadening capabilities but not addressing isolation from external actors such as humans, other robots, or machines. Although some exceptions exist (Zhu et al., 2024), it remains rare for swarms to operate alongside—let alone in support of—other systems. Yet, one of their key strengths is distributed environmental sensing: swarm robots can rapidly gather and update mission-specific data through peer-to-peer sharing (Jones et al., 2020). This swarm perception is often studied in the context of collective behavior (Brambilla et al., 2013; Trianni and Campo, 2015) and decision-making (Valentini et al., 2016b; Valentini et al., 2016a; Strobel et al., 2018; Zakir et al., 2022), but its potential to assist external agents remains underexplored (Naghsh et al., 2008; Kegeleirs et al., 2024b). Acknowledging that swarms need not be self-contained could unlock a wide range of new applications.
2.2.4 Overlooked aspects: navigation, heterogeneity, and data security
Several critical topics remain underexplored. Navigation, for instance, is often treated as an implementation detail, despite its central role in robotic behavior and its influence on experimental outcomes. Random walk is the default strategy in many studies, yet this term covers a range of behaviors—Brownian motion (Feynman et al., 2011), correlated random walk (Renshaw and Henderson, 1981), Levy walk (Zaburdaev et al., 2015), and Levy taxis (Pasternak et al., 2009)—each with different performance characteristics depending on the platform and context. For instance, a configuration optimized for Kilobots (Dimidov et al., 2016) performs poorly on E-Pucks (Kegeleirs et al., 2019). More advanced strategies such as flocking (Hauert et al., 2011; Toshiyuki et al., 2016; Baumann et al., 2022b; Brandstätter et al., 2024) and connected locomotion (Mamei et al., 2004; O’Grady et al., 2009; Slavkov et al., 2018; Carrillo-Zapata et al., 2019) are promising alternatives and deserve further attention.
Finally, secure data storage and sharing (Hunt and Hauert, 2020) remains overlooked in robot swarms. Their decentralized nature provides inherent advantages: sensitive data is fragmented, stored locally, and often shared as processed outputs rather than raw streams—all of which reduce vulnerability to unauthorized access. Still, swarms are susceptible to attacks, including infiltration by byzantine robots (Strobel et al., 2023) or physical capture of units to access onboard data. Practical deployments rely on centralized infrastructure, introducing additional risks when interfacing with external systems. Blockchain-based solutions have been proposed (Dorigo et al., 2024), but robust, field-tested security mechanisms for swarms remain an open challenge.
2.3 Key enablers for real-world deployment
2.3.1 Bridging the deployment gap
From a practical standpoint, swarm robotics research must align more closely with real-world conditions. First, providing evidence that a robot swarm can bridge the deployment gap should become standard practice. It remains uncertain whether findings obtained on current research platforms are transferable to the advanced robots required for real-world applications. In particular, it is unclear whether artificial evolution could effectively generate behaviors for more capable, sophisticated swarms. Hence, simulation experiments should be systematically validated by real-robot experiments or, at a minimum, in pseudo-reality (Ligot and Birattari, 2020; Ligot and Birattari, 2022). In addition, hardware-in-the-loop approaches can also yield valuable insights and strengthen the connection between simulation and reality (Zhang et al., 2020; Khaliq et al., 2021; Jiang and Patil, 2022). Automatic modular design approaches like AutoMoDe (Birattari et al., 2019; 2021) have shown promise in narrowing the deployment gap (Francesca et al., 2014; Kegeleirs et al., 2024c). Another promising strategy is to use a smaller, less powerful platform as a proxy for a more advanced one (Kegeleirs et al., 2024c). If control software can transfer between the two, the smaller platform can be used for large-scale testing—albeit with limited capabilities.
2.3.2 Modernizing swarm platforms
Second, standard research platforms in swarm robotics should evolve toward modern, more capable robotic systems. Hardware for sensing, actuation, and computation has become increasingly compact and affordable. Additionally, adopting standard frameworks—such as ROS—or developing alternatives (Baumann and Martinoli, 2021) would enhance robot capabilities, encourage benchmarking, and improve reusability of research outputs. The long-standing image of swarm robots as simplistic, near-useless individuals—once a powerful metaphor for emergent intelligence—is now becoming a liability. Even without competing with industrial systems, designing research-dedicated robots with similar capabilities would support emerging technologies like swarm SLAM and enable the integration of computer vision. More capable platforms would also allow researchers to design more complex missions that demand richer and more relevant behaviors. In particular, robot-to-robot communication should be more prominent in experiments to fully leverage collective intelligence. Collaboration with industry could help align platform design with real-world needs while contributing valuable expertise and technical resources.
Moreover, the concept of robot swarms can extend to systems beyond the traditional focus on mobile robots—whether terrestrial, aerial, or aquatic. Stationary systems—including intelligent structures, embedded objects, or even non-autonomous robots like wearables—can also operate as swarms. For example, a swarm of smart solar panels could use self-organization and self-assembly to maximize energy production. Recent studies also envision swarms of intelligent objects in artistic and architectural applications (Alhafnawi et al., 2021), or to enhance human-swarm interaction in activities such as brainstorming and opinion gathering (Alhafnawi et al., 2022).
2.3.3 Breaking swarm stereotypes
Conceptually, the idealized definition and role of robot swarms should be re-examined to better leverage their unique strengths. Even systems that only partially conform to traditional swarm constraints can offer significant value. A semi-autonomous swarm guided by a leader—be it a centralized system or another robot—can still exhibit self-organization, redundancy, and local interactions at the agent level. Such swarms may be ideally suited for environments like warehouses or monitoring systems. Recent research on ad hoc hierarchical structures emerging through self-organization offer another promising direction, closer to swarm definitions (Mathews et al., 2017; Zhang et al., 2023).
The expected level of flexibility in swarms may also warrant reconsideration. While adaptability is a clear strength—especially in unknown environments—many real-world applications involve (semi-)structured settings where adapting the environment to suit the swarm may be more practical. Indeed, adapting the environment to accommodate robotic systems is already common in domains such as social and assistive robotics (Šabanović, 2010; Kyrarini et al., 2021; Tsunoda and Premachandra, 2021; Kodate, 2023; Yoshikawa, 2024). In light of these shifting assumptions and hybrid designs, researchers should be more deliberate in defining the properties they expect from swarm systems and should more frequently employ formal verification methods. To support this, the development of standardized metrics and evaluation frameworks is particularly important (Ferreira Cruz et al., 2021; Kwa et al., 2023a; Milner et al., 2023).
2.3.4 Rethinking swarm’s role
Expanding beyond traditional swarm structures—through heterogeneity or support for external systems—could greatly broaden the field’s applications. In particular, leveraging swarm perception to collect and relay data for other systems is highly promising. A swarm can act as a distributed sensor network, continuously collecting and updating mission-specific environmental data. Although such data is usually used internally to refine collective behavior, it can also be viewed as a shared, dynamic environmental database. Providing this information to external systems—for instance, through communication with a separate agent—could supply critical data for completing other tasks. For instance, a robot swarm may not be ideal as a standalone search-and-rescue solution, but could still play a vital role by supporting human rescuers.
Human-swarm interaction already partially explored this concept (Kolling et al., 2016; Hussein and Abbass, 2018), enabling operators to use swarm-generated information to improve safety and efficiency. For example, a rescuer could locate victims based on swarm data, or a speleologist might rely on swarm-generated maps to plan an exploration.
Finally, swarm SLAM holds strong potential for supporting external systems. While swarms typically excel at creating coarse, abstract maps—less useful within the swarm—they are ideal for scouting missions where the swarm’s objective is to quickly relay basic mapping information to another system. Such maps can provide valuable navigational support for other robots or situational awareness for human operators, underscoring swarm SLAM’s role in exploration and reconnaissance.
Ultimately, realistic applications will require swarm robotics to integrate modern technologies and rethink some of its core assumptions. Combined with SLAM, tracking individuals across large spaces is one particularly promising use case. Multi-target, multi-camera tracking (MTMCT) and person re-identification (Re-ID) remain challenging, especially in uncontrolled environments (Amosa et al., 2023; Tang et al., 2017; Ristani and Tomasi, 2018; Gaikwad and Karmakar, 2021; Ye et al., 2022). Robot swarms offer a unique advantage: they can reposition themselves to overcome occlusions and capture richer visual data, in particular in unknown environments where strategies dependent on fixed sensor placements or path planning are impractical (Robin and Lacroix, 2016). They can also share data in real time to maintain robust identification and localization, enabling support for other robots’ navigation. For example, a hospital delivery robot could use swarm-generated data to locate the requesting doctor. In surveillance or crowd monitoring, swarms could cover blind spots left by fixed infrastructure. Crucially, swarm robots can do more than detect—they can act. Unlike passive systems, swarms can initiate local responses—either mitigating an issue until human intervention arrives or resolving it autonomously.
3 Conclusion
In this paper, we have reviewed the key challenges that currently limit the adoption of swarm robotics in real-world applications (see Table 1). In particular, progress is hindered by the lack of affordable, modern research platforms and by a rigid adherence to conventional definitions of swarm robotics. As a result, many swarm experiments remain overly simplistic and offer limited guarantees of reproducibility on real or more sophisticated robotic systems. There is also a tendency to overestimate swarm properties without sufficient empirical validation. We argue that developing reliable, modern platforms—potentially through industry collaboration—would empower researchers to perform more realistic and impactful experiments, accelerating progress in navigation, vision, and communication. We further contend that rethinking the conceptual foundations of swarm robotics could open up novel application domains. In particular, loosening strict adherence to traditional swarm principles could significantly broaden the scope of the field. For example, developing semi-autonomous swarms or swarms designed to support external systems offers promising pathways toward real-world deployment.
TABLE 1
| Category | Barrier | Key enabler |
|---|---|---|
| Practical | Outdated platforms with limited sensing, actuation, and computation | Develop modern research platforms with enhanced sensors and computing capabilities |
| Simulator limitations and deployment gap | Apply pseudo-reality testing, hardware-in-the-loop validation, and platform generalization techniques | |
| Poor integration of SLAM, vision, and communication | Embed advanced SLAM, vision, and communication stacks in new standard platforms | |
| Regulatory, ethical, and trust-related concerns | Promote transparency, human-swarm trust, and early engagement with regulators | |
| Conceptual | Rigid adherence to canonical swarm properties | Rethink the paradigm: allow hybrid or leader-guided designs while preserving decentralization |
| Unverified assumptions about swarm properties | Introduce formal validation, empirical testing, and standardized performance metrics | |
| Isolationist mindset (self-contained swarms only) | Reposition swarms as task enablers or data providers within broader multi-agent systems | |
| Overlooked aspects (e.g., navigation strategies, heterogeneity, security) | Prioritize these topics to enable richer, more realistic applications and robust deployments |
Practical and conceptual barriers in swarm robotics, and their corresponding enablers.
Statements
Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.
Author contributions
MK: Writing – original draft, Writing – review and editing. MB: Supervision, Writing – review and editing.
Funding
The author(s) declare that financial support was received for the research and/or publication of this article. The project has received funding from Belgium’s Wallonia-Brussels Federation through a ARC Advanced Project 2020 (Guaranteed by Optimization). MB acknowledges support from the Belgian Fonds de la Recherche Scientifique-FNRS, of which he is a Research Director.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declare that Generative AI was used in the creation of this manuscript. Generative AI was used to improve the quality of the language during the editing process. It was not used to generate a first draft or to conduct research.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Footnotes
1.^These missions are extensively cited in the literature (Brambilla et al., 2013; Bayındır, 2016; Nedjah and Silva Junior, 2019; Schranz et al., 2020; Calderón-Arce et al., 2022), with foraging receiving special attention (Lu et al., 2020).
References
1
AlhafnawiM.HauertS.O’DowdP. (2021). Self-organised saliency detection and representation in robot swarms. IEEE Robotics Automation Lett.6, 1487–1494. 10.1109/LRA.2021.3057567
2
AlhafnawiM.HuntE. R.LemaignanS.O’DowdP.HauertS. (2022). “Mosaix: a swarm of robot tiles for social human-swarm interaction,” in Icra 2022 (IEEE), 6882–6888. 10.1109/ICRA46639.2022.9811723
3
AllenJ. M.JoyceR.MillardA. G.GrayI. (2020). The Pi-puck ecosystem: hardware and software support for the e-puck and e-puck2, Ants 2020, eds. DorigoM.StützleT.BlesaM. J.BlumC.HamannH.HeinrichM. K.et al (ChamSwitzerland: Springer), vol. 12421, 243–255. 10.1007/978-3-030-60376-2_19
4
AmosaT. I.SebastianP.IzharL. I.IbrahimO.AyinlaL. S.BahashwanA. A.et al (2023). Multi-camera multi-object tracking: a review of current trends and future advances. Neurocomputing552, 126558. 10.1016/j.neucom.2023.126558
5
ArregiM. O.SeccoE. L. (2023). Operative guide for 4-wheel Summit XL mobile robot set-up. Acta Sci. Comput. Sci.5.
6
Australian Civil Aviation Safety Authority (2024). Applying for approval to conduct one-to-many rpa operations. Available online at: https://www.casa.gov.au/sites/default/files/2024-05/annex-b-advisory-circular-101-01-applying-approval-conduct-many-rpa-operations.pdf.
7
BaumannC.BirchH.MartinoliA. (2022a). “Leveraging multi-level modelling to automatically design behavioral arbitrators in robotic controllers,” in Iros 2022 (IEEE), 9318–9325. 10.1109/IROS47612.2022.9981817
8
BaumannC.MartinoliA. (2021). A modular functional framework for the design and evaluation of multi-robot navigation. Robotics Aut. Syst.144, 103849. 10.1016/j.robot.2021.103849
9
BaumannC.PeroliniJ.TourkiE.MartinoliA. (2022b). “Hybrid flock-formation control algorithms,” in Distributed autonomous robotic systems (Springer), 523–536. 10.1007/978-3-031-51497-5_37
10
BayındırL. (2016). A review of swarm robotics tasks. Neurocomputing172, 292–321. 10.1016/j.neucom.2015.05.116
11
BerlingerF.GauciM.NagpalR. (2021). Implicit coordination for 3D underwater collective behaviors in a fish-inspired robot swarm. Sci. Robot.6, eabd8668. 10.1126/scirobotics.abd8668
12
BirattariM.LigotA.BozhinoskiD.BrambillaM.FrancescaG.GarattoniL.et al (2019). Automatic off-line design of robot swarms: a manifesto. Front. Robot. AI6, 59. 10.3389/frobt.2019.00059
13
BirattariM.LigotA.FrancescaG. (2021). “AutoMoDe: a modular approach to the automatic off-line design and fine-tuning of control software for robot swarms,” in Automated design of machine learning and search algorithms. Editors PillayN.QuR. (ChamSwitzerland: Springer), 73–90. NCS. 10.1007/978-3-030-72069-8_5
14
BirattariM.LigotA.HasselmannK. (2020). Disentangling automatic and semi-automatic approaches to the optimization-based design of control software for robot swarms. Nat. Mach. Intell.2, 494–499. 10.1038/s42256-020-0215-0
15
BragaR. G.VaradharajanV. S.BeltrameG.St-OngeD. (2024). Swarming out of the lab: comparing relative localization methods for collective behavior. ANTS 20243, 181–194. 10.1007/978-3-031-70932-6_14
16
BrambillaM.BrutschyA.DorigoM.BirattariM. (2014). Property-driven design for swarm robotics: a design method based on prescriptive modeling and model checking. ACM Tran. Auton. Adap.9 (17), 1–17. 10.1145/2700318
17
BrambillaM.FerranteE.BirattariM.DorigoM. (2013). Swarm robotics: a review from the swarm engineering perspective. Swarm Intell.7, 1–41. 10.1007/s11721-012-0075-2
18
BrandstätterA.SmolkaS. A.StollerS. D.TiwariA.GrosuR. (2024). “Flock-formation control of multi-agent systems using imperfect relative distance measurements,” in Icra 2024 (ieee), 12193–12200. 10.1109/ICRA57147.2024.10610147
19
BredecheN.HaasdijkE.PrietoA. (2018). Embodied evolution in collective robotics: a review. Front. Robot. AI5, 12. 10.3389/frobt.2018.00012
20
Calderón-ArceC.Brenes-TorresJ. C.Solis-OrtegaR. (2022). Swarm robotics: simulators, platforms and applications review. Computation10, 80. 10.3390/computation10060080
21
CambierN.FerranteE. (2022). “AutoMoDe-Pomodoro: an evolutionary class of modular designs,” in Gecco 2022. Editor FieldsendJ. E.New York, NY: ACM, 100–103. 10.1145/3520304.3529031
22
Carrillo-ZapataD.MilnerE.HirdJ.TzoumasG.VardanegaP. J.SooriyabandaraM.et al (2020). Mutual shaping in swarm robotics: user studies in fire and rescue, storage organization, and bridge inspection. Front. Robot. AI7, 53. 10.3389/frobt.2020.00053
23
Carrillo-ZapataD.SharpeJ.WinfieldA. F. T.GiuggioliL.HauertS. (2019). Toward controllable morphogenesis in large robot swarms. IEEE Robotics Automation Lett.4, 3386–3393. 10.1109/LRA.2019.2926961
24
ChenJ.GauciM.LiW.KollingA.GroßR. (2015). Occlusion-based cooperative transport with a swarm of miniature mobile robots. IEEE Trans. Robotics31, 307–321. 10.1109/TRO.2015.2400731
25
CheraghiA. R.ShahzadS.GraffiK. (2022). “Past, present, and future of swarm robotics,” Intelligent Syst. Appl. Proc. 2021 Intelligent Syst. Conf. (IntelliSys). Springer, 3, 190–233. 10.1007/978-3-030-82199-9_13
26
CianciC. M.RaemyX.PughJ.MartinoliA. (2006). “Communication in a swarm of miniature robots: the e-puck as an educational tool for swarm robotics,” in Sab 2006. Editors NolfiS.BaldassarreG.CalabrettaR.HallamJ.MaroccoD.MeyerJ.-A.et al (LNCS: BerlinGermany: Springer), 103–115. 10.1007/978-3-540-71541-2_7
27
Code of Federal Regulations (2025). Title107.35. Available online at: https://www.ecfr.gov/current/title-14/chapter-I/subchapter-F/part-107/subpart-B/section-107.35.
28
ConnorJ.ChampionB.JoordensM. A. (2020). Current algorithms, communication methods and designs for underwater swarm robotics: a review. IEEE Sensors J.21, 153–169. 10.1109/JSEN.2020.3013265
29
CoppolaM.GuoJ.GillE.de CroonG. C. (2019). Provable self-organizing pattern formation by a swarm of robots with limited knowledge. Swarm Intell.13, 59–94. 10.1007/s11721-019-00163-0
30
CorrellN.MartinoliA. (2007). “Modeling and optimization of a swarm-intelligent inspection system,” in Distributed autonomous robotic systems (Springer), 6, 369–378. 10.1007/978-4-431-35873-2_36
31
CorrellN.MartinoliA. (2011). Modeling and designing self-organized aggregation in a swarm of miniature robots. Int. J. Robotics Res.30, 615–626. 10.1177/0278364911403017
32
DiasP. G. F.SilvaM. C.Rocha FilhoG. P.VargasP. A.CotaL. P.PessinG. (2021). Swarm robotics: a perspective on the latest reviewed concepts and applications. Sensors21, 2062. 10.3390/s21062062
33
Di CaroG. A.DucatelleF.GambardellaL. M. (2005). AntHocNet: an adaptive nature-inspired algorithm for routing in mobile ad hoc networks. Eur. Trans. Telecommun.16, 443–455. 10.1002/ett.1062
34
DimidovC.OrioloG.TrianniV. (2016). “Random walks in swarm robotics: an experiment with Kilobots,” in Ants 2016. Editors DorigoM.BirattariM.LiX.López-IbáñezM.OhkuraK.StützleT.et al (ChamSwitzerland: Springer), 9882, 185–196. 10.1007/978-3-319-44427-7_16
35
Divband SooratiM.GerdingE. H.MarchioniE.NaumovP.NormanT. J.RamchurnS. D.et al (2022). From intelligent agents to trustworthy human-centred multiagent systems. AI Commun.35, 443–457. 10.3233/AIC-220127
36
Divband SooratiM.HamannH. (2015). “The effect of fitness function design on performance in evolutionary robotics: the influence of a priori knowledge,” in Gecco 2015. Editor SilvaS. (New York, NY: ACM), 153–160. 10.1145/2739480.2754676
37
DixonC.WinfieldA. F.FisherM.ZengC. (2012). Towards temporal verification of swarm robotic systems. Robot. Auton. Syst.60, 1429–1441. 10.1016/j.robot.2012.03.003
38
DorigoM.BirattariM.BrambillaM. (2014). Swarm robotics. Scholarpedia9, 1463. 10.4249/scholarpedia.1463
39
DorigoM.FloreanoD.GambardellaL. M.MondadaF.NolfiS.BaabouraT.et al (2013). Swarmanoid: a novel concept for the study of heterogeneous robotic swarms. IEEE Robotics and Automation Mag.20, 60–71. 10.1109/MRA.2013.2252996
40
DorigoM.PachecoA.ReinaA.StrobelV. (2024). Blockchain technology for mobile multi-robot systems. Nat. Rev. Electr. Eng.1, 264–274. 10.1038/s44287-024-00034-9
41
DorigoM.TheraulazG.TrianniV. (2021). Swarm robotics: past, present, and future [point of view]. Proc. IEEE109, 1152–1165. 10.1109/JPROC.2021.3072740
42
DucatelleF.Di CaroG. A.FörsterA.BonaniM.DorigoM.MagnenatS.et al (2014). Cooperative navigation in robotic swarms. Swarm Intell.8, 1–33. 10.1007/s11721-013-0089-4
43
DucatelleF.Di CaroG. A.PinciroliC.GambardellaL. M. (2011). Self-organized cooperation between robotic swarms. Swarm Intell.5, 73–96. 10.1007/s11721-011-0053-0
44
EndoW.BaumannC.AsamaH.MartinoliA. (2023). Automatic multi-robot control design and optimization leveraging multi-level modeling: an exploration case study. IFAC-PapersOnLine56, 11462–11469. 10.1016/j.ifacol.2023.10.435
45
FdhilaR.HamdaniT. M.AlimiA. M. (2012). “A multi objective particles swarm optimization algorithm for solving the routing pico-satellites problem,” in Smc 2012 (IEEE), 1402–1407. 10.1109/ICSMC.2012.6377930
46
Ferreira CruzD. P.MaiaR. D.de CastroL. N. (2021). A framework for the analysis and synthesis of swarm intelligence algorithms. J. Exp. and Theor. Artif. Intell.33, 659–681. 10.1080/0952813X.2020.1764635
47
FeynmanR. P.LeightonR. B.SandsM. L. (2011). “The Feynman lectures on physics. Volume 1: mainly mechanics, radiation, and heat,”. New York, NY: Basic Books. millenium edn.
48
FloreanoD.HusbandsP.NolfiS. (2008). “Evolutionary robotics,” in Springer handbook of robotics. Editors SicilianoB.KhatibO. (BerlinGermany: Springer), 1423–1451. 10.1007/978-3-540-30301-5_62
49
FrancescaG.BirattariM. (2016). Automatic design of robot swarms: achievements and challenges. Front. Robot. AI3, 1–9. 10.3389/frobt.2016.00029
50
FrancescaG.BrambillaM.BrutschyA.GarattoniL.MiletitchR.PodevijnG.et al (2015). AutoMoDe-Chocolate: automatic design of control software for robot swarms. Swarm Intell.9, 125–152. 10.1007/s11721-015-0107-9
51
FrancescaG.BrambillaM.BrutschyA.TrianniV.BirattariM. (2014). AutoMoDe: a novel approach to the automatic design of control software for robot swarms. Swarm Intell.8, 89–112. 10.1007/s11721-014-0092-4
52
FujisawaR.DobataS.SugawaraK.MatsunoF. (2014). Designing pheromone communication in swarm robotics: group foraging behavior mediated by chemical substance. Swarm Intell.8, 227–246. 10.1007/s11721-014-0097-z
53
GaikwadB.KarmakarA. (2021). Smart surveillance system for real-time multi-person multi-camera tracking at the edge. J. Real-Time Image Process.18, 1993–2007. 10.1007/s11554-020-01066-8
54
Garzón RamosD.BirattariM. (2020). Automatic design of collective behaviors for robots that can display and perceive colors. Appl. Sci.10, 4654. 10.3390/app10134654
55
Garzón RamosD.HauertS. (2024). Designing robot swarms: a puzzle, a problem, and a mess. Available online at: https://arxiv.org/abs/2410.22478.
56
GharbiI.KucklingJ.Garzón RamosD.BirattariM. (2023). “Show me what you want: inverse reinforcement learning to automatically design robot swarms by demonstration,” in Icra 2023 (Piscataway, NJ, USA), 5063–5070. 10.1109/ICRA48891.2023.10160947
57
GiernackiW.SkwierczyńskiM.WitwickiW.WrońskiP.KozierskiP. (2017). “Crazyflie 2.0 quadrotor as a platform for research and education in robotics and control engineering,” in Mmar 2017 (ieee), 37–42. 10.1109/MMAR.2017.8046794
58
GjondrekajE.LoretiM.PuglieseR.TiezziF.PinciroliC.BrambillaM.et al (2012). “Towards a formal verification methodology for collective robotic systems,” in International conference on formal engineering methods (Springer), 14, 54–70. 10.1007/978-3-642-34281-3_7
59
HaasdijkE.BredecheN.EibenA. (2014). Combining environment-driven adaptation and task-driven optimisation in evolutionary robotics. PLOS ONE9, e98466. 10.1371/journal.pone.0098466
60
HamannH. (2010). Space-time continuous models of swarm robotic systems: supporting global-to-local programming. BerlinGermany: Springer. 10.1007/978-3-642-13377-0
61
HamannH.WörnH. (2008). A framework of space–time continuous models for algorithm design in swarm robotics. Swarm Intell.2, 209–239. 10.1007/s11721-008-0015-3
62
HarwellJ.GiniM. (2018). “Broadening applicability of swarm-robotic foraging through constraint relaxation,” in 2018 IEEE international conference on simulation, modeling, and programming for autonomous robots (SIMPAR) (Germany: IEEE), 116–122. 10.1109/SIMPAR.2018.8376280
63
HasselmannK.BirattariM. (2020). Modular automatic design of collective behaviors for robots endowed with local communication capabilities. PeerJ Comput. Sci.6, e291. 10.7717/peerj-cs.291
64
HasselmannK.LigotA.BirattariM. (2023). Automatic modular design of robot swarms based on repertoires of behaviors generated via novelty search. Swarm Evol. Comput.83, 101395. 10.1016/j.swevo.2023.101395
65
HasselmannK.LigotA.RuddickJ.BirattariM. (2021). Empirical assessment and comparison of neuro-evolutionary methods for the automatic off-line design of robot swarms. Nat. Commun.12, 4345. 10.1038/s41467-021-24642-3
66
HauertS.BhatiaS. N. (2014). Mechanisms of cooperation in cancer nanomedicine: towards systems nanotechnology. Trends Biotechnol.32, 448–455. 10.1016/j.tibtech.2014.06.010
67
HauertS.LevenS.VargaM.RuiniF.CangelosiA.ZuffereyJ.-C.et al (2011). Reynolds flocking in reality with fixed-wing robots: communication range vs. maximum turning rate. IROS 2011 (IEEE), 5015–5020. 10.1109/IROS.2011.6095129
68
HeckerJ. P.MosesM. E. (2015). Beyond pheromones: evolving error-tolerant, flexible, and scalable ant-inspired robot swarms. Swarm Intell.9, 43–70. 10.1007/s11721-015-0104-z
69
HuntE. R.HauertS. (2020). A checklist for safe robot swarms. Nat. Mach. Intell.2, 420–422. 10.1038/s42256-020-0213-2
70
HuntE. R.JonesS.HauertS. (2019). Testing the limits of pheromone stigmergy in high-density robot swarms. R. Soc. open Sci.6, 190225. 10.1098/rsos.190225
71
HusseinA.AbbassH. (2018). “Mixed initiative systems for human-swarm interaction: opportunities and challenges,” in 2018 2nd annual systems modelling conference (SMC) (IEEE), 1–8. 10.1109/SYSMC.2018.8509744
72
HusseinA.ElsawahS.AbbassH. A. (2020). The reliability and transparency bases of trust in human-swarm interaction: principles and implications. Ergonomics63, 1116–1132. 10.1080/00140139.2020.1764112
73
HüttenrauchM.ŠošićA.NeumannG. (2017). Guided deep reinforcement learning for swarm systems. Available online at: https://arxiv.org/abs/1709.06011.
74
HüttenrauchM.ŠošićA.NeumannG. (2019). Deep reinforcement learning for swarm systems. J. Mach. Learn. Res.20, 1–31.
75
JakobiN.HusbandsP.HarveyI. (1995). “Noise and the reality gap: the use of simulation in evolutionary robotics,” in Ecal ’95. Editors MoránF.MorenoA.MereloJ. J.ChacónP. (LNAI: Springer), 929, 704–720. 10.1007/3-540-59496-5_337
76
JiangZ.PatilT. (2022). “A distributed hardware-in-the-loop simulation testbed for swarms of small autonomous vehicles,” in Aiaa 2022 (American Institute of Aeronautics and Astronautics), 4057. 10.2514/6.2022-4057
77
Jimenez RomeroC.YegenogluA.Pérez MartínA.Diaz-PierS.MorrisonA. (2024). Emergent communication enhances foraging behavior in evolved swarms controlled by spiking neural networks. Swarm Intell.18, 1–29. 10.1007/s11721-023-00231-6
78
JonesS.MilnerE.SooriyabandaraM.HauertS. (2020). Distributed situational awareness in robot swarms. Adv. Intell. Syst.2, 2000110. 10.1002/aisy.202000110
79
JonesS.WinfieldA.HauertS.StudleyM. (2019). Onboard evolution of understandable swarm behaviors. Adv. Intell. Syst.1, 1900031. 10.1002/aisy.201900031
80
KaiserT. K.HamannH. (2022). Innate motivation for robot swarms by minimizing surprise: from simple simulations to real-world experiments. IEEE Trans. Robotics38, 3582–3601. 10.1109/TRO.2022.3181004
81
KegeleirsM.Garzón RamosD.BirattariM. (2019). “Random walk exploration for swarm mapping,” in Taros 2019. Editors AlthoeferK.KonstantinovaJ.ZhangK. (ChamSwitzerland: Springer), 11650, 211–222. 10.1007/978-3-030-25332-5_19
82
KegeleirsM.Garzón RamosD.BirattariM. (2025). Deimos: a ros-ready operating system for the e-puck. J. Open Res. Softw.13, 3. 10.5334/jors.437
83
KegeleirsM.Garzón RamosD.HasselmannK.GarattoniL.FrancescaG.BirattariM. (2024c). Transferability in the automatic off-line design of robot swarms: from sim-to-real to embodiment and design-method transfer across different platforms. IEEE Robotics Automation Lett.9, 2758–2765. 10.1109/LRA.2024.3360013
84
KegeleirsM.Garzón RamosD.Legarda HerranzG.GharbiI.SzpirerJ.DebeirO.et al (2024a). Collective perception for tracking people with a robot swarm. Available online at: https://arxiv.org/abs/2410.06720.
85
KegeleirsM.Garzón RamosD.Legarda HerranzG.GharbiI.SzpirerJ.HasselmannK.et al (2024b). Leveraging swarm capabilities to assist other systems. Available online at: https://arxiv.org/abs/2405.04079.
86
KegeleirsM.GrisettiG.BirattariM. (2021). Swarm SLAM: challenges and perspectives. Front. Robot. AI8, 618268. 10.3389/frobt.2021.618268
87
KegeleirsM.TodescoR.Garzón RamosD.Legarda HerranzG.BirattariM. (2022). Mercator: hardware and software architecture for experiments in swarm SLAM. Tech. Rep. TR/IRIDIA/2022-012, IRIDIA, ULB.
88
KhaliqA. A.Di RoccoM.SaffiottiA. (2014). Stigmergic algorithms for multiple minimalistic robots on an rfid floor. Swarm Intell.8, 199–225. 10.1007/s11721-014-0096-0
89
KhaliqS.AhsanS.NisarM. D. (2021). “Multi-platform hardware in the loop (hil) simulation for decentralized swarm communication using ros and gazebo,” in International symposium on a world of wireless, Mobile and multimedia networks (WoWMoM) (IEEE), 310–315. 10.1109/WoWMoM51794.2021.00057
90
KodateN. (2023). Why does implementation of innovative technologies matter for care professionals?Available online at: http://ffj.ehess.fr/index/article/433/why-does-implementation-of-innovative-technologies-matter-for-care-professionals.html.
91
KoenigN. P.HowardA. (2004). Design and use paradigms for Gazebo, an open-source multi-robot simulator. IROS 2004 Piscataw. IEEE3, 2149–2154. 10.1109/IROS.2004.1389727
92
KollingA.WalkerP.ChakrabortyN.SycaraK.LewisM. (2016). Human interaction with robot swarms: a survey. IEEE Trans. Human-Machine Syst.46, 9–26. 10.1109/THMS.2015.2480801
93
KonurS.DixonC.FisherM. (2012). Analysing robot swarm behaviour via probabilistic model checking. Robot. Auton. Syst.60, 199–213. 10.1016/j.robot.2011.10.005
94
KoosS.MouretJ.-B.DoncieuxS. (2013). The transferability approach: crossing the reality gap in evolutionary robotics. IEEE Trans. Evol. Comput.17, 122–145. 10.1109/TEVC.2012.2185849
95
KucklingJ. (2023). Recent trends in robot learning and evolution for swarm robotics. Front. Robot. AI10, 1134841. 10.3389/frobt.2023.1134841
96
KucklingJ.LuckeyR.AvrutinV.VardyA.ReinaA.HamannH. (2024). “Do we run large-scale multi-robot systems on the edge? more evidence for two-phase performance in system size scaling,” in Icra 2024 (ieee), 4562–4568. 10.1109/ICRA57147.2024.10610771
97
KucklingJ.van PeltV.BirattariM. (2022). AutoMoDe-Cedrata: automatic design of behavior trees for controlling a swarm of robots with communication capabilities. SN Comput. Sci.3, 136. 10.1007/s42979-021-00988-9
98
KwaH. L.KitJ. L.HorsevadN.PhilippotJ.SavariM.BouffanaisR. (2023a). Adaptivity: a path towards general swarm intelligence?Front. Robot. AI10, 1163185. 10.3389/frobt.2023.1163185
99
KwaH. L.PhilippotJ.BouffanaisR. (2023b). Effect of swarm density on collective tracking performance. Swarm Intell.17, 253–281. 10.1007/s11721-023-00225-4
100
KwaH. L.TokićG.BouffanaisR.YueD. K. (2020). Heterogeneous swarms for maritime dynamic target search and tracking. Glob. Oceans 2020 (IEEE)4, 1–8. 10.1109/IEEECONF38699.2020.9389145
101
KyrariniM.LygerakisF.RajavenkatanarayananA.SevastopoulosC.NambiappanH. R.ChaitanyaK. K.et al (2021). A survey of robots in healthcare. Technologies9, 8. 10.3390/technologies9010008
102
LajoieP.-Y.BeltrameG. (2023). Swarm-SLAM: sparse decentralized collaborative simultaneous localization and mapping framework for multi-robot systems. IEEE Robotics Automation Lett.9, 475–482. 10.1109/LRA.2023.3333742
103
LawJ.YuJ.TangW.GongZ.WangX.SunY. (2023). Micro/nanorobotic swarms: from fundamentals to functionalities. ACS nano17, 12971–12999. 10.1021/acsnano.2c11733
104
LeofanteF.SchuppS.AbrahamE.TacchellaA. (2019). “Engineering controllers for swarm robotics via reachability analysis in hybrid systems,” in European conference on modelling and simulation (Springer), 407–413. 10.7148/2019-0407
105
LigotA.BirattariM. (2020). Simulation-only experiments to mimic the effects of the reality gap in the automatic design of robot swarms. Swarm Intell.14, 1–24. 10.1007/s11721-019-00175-w
106
LigotA.HasselmannK.BirattariM. (2020). AutoMoDe-Arlequin: neural networks as behavioral modules for the automatic design of probabilistic finite state machines, Ants 2020, eds. DorigoM.StützleT.BlesaM. J.BlumC.HamannH.HeinrichM. K.et al (ChamSwitzerland: Springer), vol. 12421, 271–281. 10.1007/978-3-030-60376-2_21
107
LigotA.BirattariM. (2022). On using simulation to predict the performance of robot swarms. Sci. Data9, 788. 10.1038/s41597-022-01895-1
108
LiuR.CaiZ.LewisM.LyonsJ.SycaraK. (2019). “Trust repair in human-swarm teams+,” in RO-MAN 2019 (IEEE), 1–6. 10.1109/RO-MAN46459.2019.8956420
109
LuQ.FrickeG. M.EricksenJ. C.MosesM. E. (2020). Swarm foraging review: closing the gap between proof and practice. Curr. Robot. Rep.1, 215–225. 10.1007/s43154-020-00018-1
110
LyonsJ. B.CapiolaA.AdamsJ. A.MatorJ. D.CherryE.BarreraK. (2025). Examining the human-centred challenges of human–swarm interaction. Philos. Trans. A383, 20240140. 10.1098/rsta.2024.0140
111
MameiM.VasiraniM.ZambonelliF. (2004). Experiments of morphogenesis in swarms of simple mobile robots. Appl. Artif. Intell.18, 903–919. 10.1080/08839510490509081
112
MassinkM.BrambillaM.LatellaD.DorigoM.BirattariM. (2013). On the use of bio-pepa for modelling and analysing collective behaviours in swarm robotics. Swarm Intell.7, 201–228. 10.1007/s11721-013-0079-6
113
MathewsN.ChristensenA. L.O’GradyR.MondadaF.DorigoM. (2017). Mergeable nervous systems for robots. Nat. Commun.8, 439. 10.1038/s41467-017-00109-2
114
McGuireK.De WagterC.TuylsK.KappenH.de CroonG. C. (2019). Minimal navigation solution for a swarm of tiny flying robots to explore an unknown environment. Sci. Robotics4, eaaw9710. 10.1126/scirobotics.aaw9710
115
MendiburuF. J.Garzón RamosD.MoraisM. R. A.LimaA. M. N.BirattariM. (2022). AutoMoDe-Mate: automatic off-line design of spatially-organizing behaviors for robot swarms. Swarm Evol. Comput.74, 101118. 10.1016/j.swevo.2022.101118
116
MerloE.PinciroliC.PaneratiJ.FamelisM.BeltrameG. (2022). “Automated extraction and checking of property models from source code for robot swarms,” in International workshop on robotics software engineerin, 47–54. 10.1145/3526071.3527516
117
MilnerE.SooriyabandaraM.HauertS. (2023). Swarm performance indicators: metrics for robustness, fault tolerance, scalability and adaptability. Available online at: https://arxiv.org/abs/2311.01944.
118
MondadaF.BonaniM.RaemyX.PughJ.CianciC.KlaptoczA.et al (2009). “The e-puck, a robot designed for education in engineering,” in Robotica 2009. Editors GonçalvesP.TorresP.AlvesC. (Castelo Branco, Portugal: Instituto Politécnico de Castelo Branco), 59–65.
119
MondadaL.KarimM. E.MondadaF. (2016). Electroencephalography as implicit communication channel for proximal interaction between humans and robot swarms. Swarm Intell.10, 247–265. 10.1007/s11721-016-0127-0
120
NaS.QiuY.TurgutA. E.UlrichJ.KrajníkT.YueS.et al (2021). Bio-inspired artificial pheromone system for swarm robotics applications. Adapt. Behav.29, 395–415. 10.1177/1059712320918936
121
NaghshA. M.GancetJ.TanotoA.RoastC. (2008). “Analysis and design of human-robot swarm interaction in firefighting,” in RO-MAN 2008-the 17th IEEE international symposium on robot and human interactive communication (IEEE), 255–260. 10.1109/ROMAN.2008.4600675
122
NagiJ.GiustiA.GambardellaL. M.Di CaroG. A. (2014). “Human-swarm interaction using spatial gestures,” in Iros 2014 (IEEE), 3834–3841. 10.1109/IROS.2014.6943101
123
NaisehM.SooratiM. D.RamchurnS. (2024). “Outlining the design space of explainable swarm (xSwarm): experts’ perspective,” in Distributed autonomous robotic systems (Springer), 28–41. 10.1007/978-3-031-51497-5_3
124
NamC.WalkerP.LiH.LewisM.SycaraK. (2019). Models of trust in human control of swarms with varied levels of autonomy. IEEE Trans. Human-Machine Syst.50, 194–204. 10.1109/THMS.2019.2896845
125
NedjahN.Silva JuniorL. (2019). Review of methodologies and tasks in swarm robotics towards standardization. Swarm Evol. Comput.50, 100565. 10.1016/j.swevo.2019.100565
126
NolfiS. (2021). Behavioral and cognitive robotics: an adaptive perspective. RomeItaly Inst. Cognitive Sci. Technol. Res. Counc.
127
NouyanS.GroßR.BonaniM.MondadaF.DorigoM. (2009). Teamwork in self-organized robot colonies. IEEE Trans. Evol. Comput.13, 695–711. 10.1109/TEVC.2008.2011746
128
O’GradyR.ChristensenA. L.DorigoM. (2009). SWARMORPH: multirobot morphogenesis using directional self-assembly. IEEE Trans. Robotics25, 738–743. 10.1109/TRO.2008.2012341
129
OguzS.AllwrightM.ZhuW.WahbyM.GaroneE.DorigoM. (2022). S-drone: an open-source quadrotor for experimentation in swarm robotics. Tech. Rep. TR/IRIDIA/2022-010, IRIDIAUniversité libre Brux. BrusselsBelgium.
130
PasternakZ.BartumeusF.GrassoF. W. (2009). Lévy-taxis: a novel search strategy for finding odor plumes in turbulent flow-dominated environments. J. Phys. A Math. Theor.42, 434010. 10.1088/1751-8113/42/43/434010
131
PavlivM.SchianoF.ReardonC.FloreanoD.LoiannoG. (2021). Tracking and relative localization of drone swarms with a vision-based headset. IEEE Robotics Automation Lett.6, 1455–1462. 10.1109/LRA.2021.3051565
132
PinciroliC.BirattariM.TuciE.DorigoM.del Rey ZapateroM.VinkoT.et al (2008a). “Self-organizing and scalable shape formation for a swarm of pico satellites,” in AHS 2008 (NoordwijkNetherlands: ieee), 57–61. 10.1109/AHS.2008.41
133
PinciroliC.BirattariM.TuciE.DorigoM.del Rey ZapateroM.VinkoT.et al (2008b). “Lattice formation in space for a swarm of pico satellites,” in Ants 2008. Editors DorigoM.BirattariM.BlumC.ClercM.StützleT.WinfieldA. (Springer), 347–354. 10.1007/978-3-540-87527-7_36
134
PinciroliC.TrianniV.O’GradyR.PiniG.BrutschyA.BrambillaM.et al (2012). ARGoS: a modular, parallel, multi-engine simulator for multi-robot systems. Swarm Intell.6, 271–295. 10.1007/s11721-012-0072-5
135
PitonakovaL.WinfieldA.CrowderR. (2018). “Recruitment near worksites facilitates robustness of foraging e-puck swarms to global positioning noise,” in Iros 2018. Editors St-OngeD.PinciroliC.BeltrameG. (PiscatawayNJUSA: IEEE), 4276–4281. 10.1109/IROS.2018.8593788
136
PodevijnG.O’gradyR.MathewsN.GillesA.Fantini-HauwelC.DorigoM. (2016). Investigating the effect of increasing robot group sizes on the human psychophysiological state in the context of human–swarm interaction. Swarm Intell.10, 193–210. 10.1007/s11721-016-0124-3
137
PopoviciM.-A.GilI. E.MontijanoE.TardioliD. (2022). “Distributed dynamic assignment of multiple mobile targets based on person re-identification,” in Iberian robotics conference (Springer), 211–222.
138
QuraishiA.MartinoliA. (2022). “Distributed cooperative localization with efficient pairwise range measurements,” in Distributed autonomous robotic systems (Springer), 134–147. 10.1007/978-3-030-92790-5_11
139
ReinaA.CopeA. J.NikolaidisE.MarshallJ. A.SaboC. (2021). Ark: augmented reality for kilobots. IEEE Robotics Automation Lett.2, 1755–1761. 10.1109/LRA.2017.2700059
140
RenshawE.HendersonR. (1981). The correlated random walk. J. Appl. Probab.18, 403–414. 10.2307/3213286
141
RistaniE.TomasiC. (2018). “Features for multi-target multi-camera tracking and re-identification,” in 2018 IEEE/CVF conference on computer vision and pattern recognition (PiscatawayNJUSA: ieee), 6036–6046. 10.1109/CVPR.2018.00632
142
RobinC.LacroixS. (2016). Multi-robot target detection and tracking: taxonomy and survey. Auton. Robots40, 729–760. 10.1007/s10514-015-9491-7
143
RouffC.VanderbiltA.HincheyM.TruszkowskiW.RashJ. (2004). “Properties of a formal method for prediction of emergent behaviors in swarm-based systems,” in International conference on software engineering and formal methods (IEEE), 24–33. 10.1109/SEFM.2004.1347500
144
RubensteinM.AhlerC.HoffN.CabreraA.NagpalR. (2014). Kilobot: a low cost robot with scalable operations designed for collective behaviors. Robot. Auton. Syst.62, 966–975. 10.1016/j.robot.2013.08.006
145
RubensteinM.AhlerC.NagpalR. (2012). “Kilobot: a low cost scalable robot system for collective behaviors,” in Icra 2012, 3293–3298. 10.1109/ICRA.2012.6224638
146
RussellR. A. (1997). Heat trails as short-lived navigational markers for mobile robots. ICRA 1997 (IEEE)4, 3534–3539. 10.1109/ROBOT.1997.606882
147
ŠabanovićS. (2010). Robots in society, society in robots: mutual shaping of society and technology as a framework for social robot design. Int. J. Soc. Robotics2, 439–450. 10.1007/s12369-010-0066-7
148
SalmanM.Garzón RamosD.BirattariM. (2024). Automatic design of stigmergy-based behaviours for robot swarms. Commun. Eng.3, 30. 10.1038/s44172-024-00175-7
149
SalmanM.Garzón RamosD.HasselmannK.BirattariM. (2020). Phormica: photochromic pheromone release and detection system for stigmergic coordination in robot swarms. Front. Robot. AI7, 195. 10.3389/frobt.2020.591402
150
SalvatoE.FenuG.MedvetE.PellegrinoF. A. (2021). Crossing the reality gap: a survey on sim-to-real transferability of robot controllers in reinforcement learning. IEEE Access9, 153171–153187. 10.1109/access.2021.3126658
151
SchillingF.SoriaE.FloreanoD. (2022). On the scalability of vision-based drone swarms in the presence of occlusions. IEEE Access10, 28133–28146. 10.1109/ACCESS.2022.3158758
152
SchmicklT.MöslingerC.CrailsheimK. (2006). “Collective perception in a robot swarm,” in International workshop on swarm robotics. Editors ŞahinE.SpearsW. M.WinfieldA. F. T. (HeidelbergGermany: Springer), 4433, 144–157. 10.1007/978-3-540-71541-2_10
153
SchranzM.UmlauftM.SendeM.ElmenreichW. (2020). Swarm robotic behaviors and current applications. Front. Robot. AI7, 36. 10.3389/frobt.2020.00036
154
SlavkovI.Carrillo-ZapataD.CarranzaN.DiegoX.JanssonF.KaandorpJ.et al (2018). Morphogenesis in robot swarms. Sci. Robot.3, eaau9178. 10.1126/scirobotics.aau9178
155
SongY.FangX.LiuB.LiC.LiY.YangS. X. (2020). A novel foraging algorithm for swarm robotics based on virtual pheromones and neural network. Appl. Soft Comput.90, 106156. 10.1016/j.asoc.2020.106156
156
SoriaE.SchianoF.FloreanoD. (2020). “Swarmlab: a matlab drone swarm simulator,” in Iros 2020, 8005–8011. 10.1109/IROS45743.2020.9340854
157
SpaeyG.KegeleirsM.Garzón RamosD.BirattariM. (2020). Evaluation of alternative exploration schemes in the automatic modular design of robot swarms. In Artificial intelligence and machine learning: bnaic 2019, benelearn 2019, eds. BogaertsB.BontempiG.GeurtsP.HarleyN.LebichotB.LenaertsT.et alSpringer), vol. 1196 of CCIS. 18–33. 10.1007/978-3-030-65154-1_2
158
StrobelV.Castello FerrerE.DorigoM. (2018). “Managing byzantine robots via blockchain technology in a swarm robotics collective decision making scenario,” in AAMAS 2018 (RichlandSCUSA: international foundation for autonomous agents and multiagent systems (IFAAMAS)), 541–549.
159
StrobelV.PachecoA.DorigoM. (2023). Robot swarms neutralize harmful byzantine robots using a blockchain-based token economy. Sci. Robot.8, eabm4636. 10.1126/scirobotics.abm4636
160
SureshA.MartínezS. (2019). Gesture based human-swarm interactions for formation control using interpreters. IFAC-PapersOnLine51, 83–88. 10.1016/j.ifacol.2019.01.033
161
SzpirerJ.Garzón RamosD.BirattariM. (2024). “Automatic design of robot swarms that perform composite missions: an approach based on inverse reinforcement learning,” in Iros 2024 (ieee), 2467.
162
TalamaliM. S.BoseT.HaireM.XuX.MarshallJ. A.ReinaA. (2020). Sophisticated collective foraging with minimalist agents: a swarm robotics test. Swarm Intell.14, 25–56. 10.1007/s11721-019-00176-9
163
TalamaliM. S.SahaA.MarshallJ. A.ReinaA. (2021). When less is more: robot swarms adapt better to changes with constrained communication. Sci. Robotics6, eabf1416. 10.1126/scirobotics.abf1416
164
TangS.AndrilukaM.AndresB.SchieleB. (2017). “Multiple people tracking by lifted multicut and person re-identification,” in 2017 IEEE conference on computer vision and pattern recognition (CVPR) (PiscatawayNJUSA: ieee), 3539–3548. 10.1109/CVPR.2017.394
165
TaraporeD.GroßR.ZaunerK.-P. (2020). Sparse robot swarms: moving swarms to real-world applications. Front. Robot. AI7, 83. 10.3389/frobt.2020.00083
166
ToshiyukiY.NakataniS.AdachiA.KadotaM.OhkuraK. (2016). Adaptive role assignment for self-organized flocking of a real robotic swarm. Artif. Life Robotics21, 405–410. 10.1007/s10015-016-0331-4
167
TrianniV.CampoA. (2015). “Fundamental collective behaviors in swarm robotics,” in Handbook of computational intelligence. Editors KacprzykJ.PedryczW. (HeidelbergGermany: Springer), 1377–1394. 10.1007/978-3-662-43505-2
168
TrianniV.López-IbáñezM. (2015). Advantages of task-specific multi-objective optimisation in evolutionary robotics. PLOS ONE10, e0136406. 10.1371/journal.pone.0136406
169
TsunodaM.PremachandraC. (2021). Remote control of a wheeled robot by visible light for support in infectious disease hospitals. IEEE Access9, 124165–124175. 10.1109/ACCESS.2021.3110891
170
UK Civil Aviation Authority (2022). Unmanned aircraft systems: rotary wing swarm operations, visual line of sight, requirements, guidance and policy. Available online at: https://www.caa.co.uk/publication/download/18053.
171
ValentiniG.BrambillaD.HamannH.DorigoM. (2016a). “Collective perception of environmental features in a robot swarm,”. Ants 2016. Editors DorigoM.BirattariM.LiX.López-IbáñezM.OhkuraK.PinciroliC.et al (ChamSwitzerland: Springer), 9882, 65–76. 10.1007/978-3-319-44427-7_6
172
ValentiniG.FerranteE.HamannH.DorigoM. (2016b). Collective decision with 100 kilobots: speed versus accuracy in binary discrimination problems. Aut. agents multi-agent Syst.30, 553–580. 10.1007/s10458-015-9323-3
173
van DiggelenF.De CarloM.CambierN.FerranteE.EibenG. (2024). “Emergence of specialised collective behaviors in evolving heterogeneous swarms,” in International conference on parallel problem solving from nature (Springer), 4, 53–69. 10.1007/978-3-031-70068-2_4
174
VerlekarH.JoshiK. (2017). “Ant and bee inspired foraging swarm robots using computer vision,” in Iceeccot 2017 (ieee), 191–195. 10.1109/ICEECCOT.2017.8284663
175
WangJ.LiuY.NiuS.JingW.SongH. (2021). Throughput optimization in heterogeneous swarms of unmanned aircraft systems for advanced aerial mobility. IEEE Trans. Intelligent Transp. Syst.23, 2752–2761. 10.1109/TITS.2021.3082512
176
WeiC.HindriksK. V.JonkerC. M. (2016). Dynamic task allocation for multi-robot search and retrieval tasks. Appl. Intell.45, 383–401. 10.1007/s10489-016-0771-5
177
WilsonS.GlotfelterP.WangL.MayyaS.NotomistaG.MoteM.et al (2020). The robotarium: globally impactful opportunities, challenges, and lessons learned in remote-access, distributed control of multirobot systems. IEEE Control Syst. Mag.40, 26–44. 10.1109/MCS.2019.2949973
178
WinfieldA. F.SaJ.Fernández-GagoM.-C.DixonC.FisherM. (2005). On formal specification of emergent behaviours in swarm robotic systems. Int. J. Adv. Robotic Syst.2, 39. 10.5772/5769
179
WinfieldA. F.SwanaM.IvesJ.HauertS. (2025). On the ethical governance of swarm robotic systems in the real world. Philos. Trans. A383, 20240142. 10.1098/rsta.2024.0142
180
YasudaT.OhkuraK. (2019). Sharing experience for behavior generation of real swarm robot systems using deep reinforcement learning. J. Robotics Mechatronics31, 520–525. 10.20965/jrm.2019.p0520
181
YeM.ShenJ.LinG.XiangT.ShaoL.HoiS. C. H. (2022). Deep learning for person re-identification: a survey and outlook. IEEE Trans. Pattern Analysis Mach. Intell.44, 2872–2893. 10.1109/TPAMI.2021.3054775
182
YoshikawaK. (2024). Robots carrying medicine in place of humans. Available online at: https://global.toyota/en/mobility/frontier-research/40390293.html.
183
ZaburdaevV.DenisovS.KlafterJ. (2015). Lévy walks. Rev. Mod. Phys.87, 483–530. 10.1103/RevModPhys.87.483
184
ZahadatP.SchmicklT. (2016). Division of labor in a swarm of autonomous underwater robots by improved partitioning social inhibition. Adapt. Behav.24, 87–101. 10.1177/1059712316633028
185
ZakirR.DorigoM.ReinaA. (2022). “Robot swarms break decision deadlocks in collective perception through cross-inhibition,” in Ants 2022. Editors DorigoM.HamannH.López-IbáñezM.García-NietoJ.EngelbrechtA.PinciroliC.et al (ChamSwitzerland: Springer), 13491, 209–221. 10.1007/978-3-031-20176-9_17
186
ZhangY.OğuzS.WangS.GaroneE.WangX.DorigoM.et al (2023). Self-reconfigurable hierarchical frameworks for formation control of robot swarms. IEEE Trans. Cybern.54, 87–100. 10.1109/TCYB.2023.3237731
187
ZhangZ.YangW.ShiZ.ZhongY. (2020). “Hardware-in-the-loop simulation platform for unmanned aerial vehicle swarm system: architecture and application,” in Chinese control conference 2020 (IEEE), 58–64. 10.23919/CCC50068.2020.9188541
188
ZhaoW.QueraltaJ. P.WesterlundT. (2020). “Sim-to-real transfer in deep reinforcement learning for robotics: a survey,” in Symposium series on computational intelligence (IEEE), 737–744. 10.1109/SSCI47803.2020.9308468
189
ZhuW.OguzS.HeinrichM. K.AllwrightM.WahbyM.ChristensenA. L.et al (2024). Self-organizing nervous systems for robot swarms. Sci. Robot.9, eadl5161. 10.1126/scirobotics.adl5161
Summary
Keywords
swarm robotics, real-world applications, distributed systems, multi-robot systems, design methodology, robot communication
Citation
Kegeleirs M and Birattari M (2025) Towards applied swarm robotics: current limitations and enablers. Front. Robot. AI 12:1607978. doi: 10.3389/frobt.2025.1607978
Received
08 April 2025
Accepted
23 May 2025
Published
13 June 2025
Volume
12 - 2025
Edited by
Roland Bouffanais, University of Geneva, Switzerland
Reviewed by
Hian Lee Kwa, Thales, Singapore
Razanne Abu-Aisheh, University of Bristol, United Kingdom
Updates
Copyright
© 2025 Kegeleirs and Birattari.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Miquel Kegeleirs, miquel.kegeleirs@ulb.be; Mauro Birattari, mauro.birattari@ulb.be
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.