Swarm intelligence

Concept, vision and application

Thought Leadership Abstract

Swarming as an evolutionary advantage Using animal swarms as source of inspiration, we define swarm intelligence (SI) and investigate how SI can be applied in the human domain. Natural swarms demonstrate how the ability to apply the power of many through self-organization creates an evolutionary advantage, hence an enabler of survival of the species. We therefore position SI as an organizational paradigm.

Swarm intelligence as a way to exploit human capital Humans, being just a specific kind of animal, also have a tendency to swarm, but more from a social cultural than a survival perspective. Our capability for self-organization - the key feature of SI - is unprecedented. The majority of the artifacts that surround us are the result of how we mutually interact, i.e. how we manage to organize ourselves as humans.

Apart from a natural fit, today there is also a need to explore what we can do with SI. We are being challenged by an ever-growing complexity and accelerating speed of change, hence growing uncertainty and turmoil. Classical hierarchical organization models fail to deal with such an environment, while SI-based organization models provide a way out. SI brings the power of decision making to a local and distributed level, relying on the people at this level to resolve the issues they are confronted with. This, by optimally utilizing the capabilities and drivers of these individuals to organize themselves in response to the environment and to each other.

Swarm intelligence as the basis for a global platform If we want to explore how SI can help us at macroeconomic level, there are some challenges. Turning to the digital world, we see there how people adopt swarm behavior, e.g. in communities, often on a global scale. This creates tremendous opportunities to organize ourselves, regardless of place or time. However, we also see how SI-models do not really emerge as the conditions for SI to function are not met, quite often lack of trust being the major blocking factor. What is missing are the platforms for people to interact in a trusted way, relying on a next generation internet only accessible for people with e-identities issued by governmental authorities, an Internet of People. Blockchain is an example of a platform architecture that enables people to operate in a swarm manner.

Swarm intelligence taking advantage of commoditized computational devices In a similar way, as we explored for organizations, centralized expensive processing has become a bottleneck in many areas. As computing power nowadays is omnipresent, is interconnected and has become commoditized, a decentralized processing architecture can be realized at low cost. Such an architecture is also more natural than a central one, as needs for information processing and consumption reside locally. SI-principles can very well be applied to create groups of self- organizing relatively cheap devices, like drones, robots, smartphones, to implement the power of many in the automation domain at a high degree of resilience and robustness.

02 Swarm intelligence in converging digital and physical worlds To develop solutions relying on SI-principles, we need platforms, operating systems and architectures to enable a swarm-like self-organizing type of processing across a group of dispersed devices. Swarm-algorithms, as we explore, provide the layer of logic on top for devices to exhibit self-organization behavior in dealing with a particular target in a problem space. These local computational devices do not only mutually interconnect, but also provide the means for people to connect and to swarm in the digital world, regardless of place or time. As the physical human and the digital virtual world become more and more interwoven, we expect to see SI-based vertical solutions, e.g. aiming at specific groups of professionals like medical specialists, to support collaboration on common tasks, potentially at a global scale.

What can we expect in the future from swarm intelligence? Is swarm intelligence going to be a game changer? No doubt, in our view the answer is YES. In our complex world and fast-paced, constantly evolving environment, it is more than ever time to act as a person or a company smarter than others. Let’s be swarm intelligence minded to better: answer questions, make predictions, optimize opinions, generate forecasts, reach decisions.

So, in a world in which almost all people and devices are connected, we argue in this paper that we should start to think about how to exploit this huge potential by developing SI-based digital and human solutions. It’s striking to see how swarm intelligence is at stake in most of the influential technologies and associated business opportunities and challenges Atos put forward in its vision Journey 2022.

Let’s be ready to reap the rewards of the swarm intelligence promise: - It’s time to engage with nature - Will your company be a disrupter or will it be disrupted? - Take the right decisions faster than competition - Create top class value from an ultra-connected world

Swarm intelligence 03 Contents

05 What is swarm intelligence about? 06 Obtaining swarm intelligence in man-made systems 10 Is swarm intelligence going to be a game changer?

12 Conclusion

04 What is swarm intelligence about?

2.1 Defining the concept 2.2.2 Self-organization as foundation of swarm intelligence 2.1.1 Swarm behavior Within a swarm, the individual agents Swarm intelligence is derived from swarm perpetually select their next course of action behavior as found in nature. Swarm from a range of alternatives, in view of the behavior, or swarming, is defined here as actual state of affairs and the contribution of “a collective behavior exhibited by agents, their best next action to the attainment of particularly animals, of similar kind which the group objective. The heuristics that the aggregate together, milling about the same agents apply to determine their next action spot or moving en masse in some direction” are based on control loop principles, like [1.]. 2.2 What are the mechanics feedback and feedforward, to either correct leading to swarm intelligence? or enforce an agent’s actions, bringing it As human observers, swarm behavior stepwise closer to the desired state. This is in intrigues us. Take for example of a flock of 2.2.1 Elementary characteristics essence what self-organization is about; the sparrows. While “being as free as a bird”, SI does not result from just any kind of group perpetual mutual alignment of the behavior when flocking, individual sparrows voluntarily behavior. For SI to emerge, the underlying of the agents from the perspective of each give in their freedom entirely, but temporarily, group behavior needs to have the following one’s contribution to the group objective. to become part of something bigger, i.e. the elementary characteristics: swarm. Research has shown that this type To demonstrate these principles of self- of swarming is an effective strategy against SI involves a group of similar or equivalent organization, we can look at ants foraging predators, like falcons [3.]. So, by swarming agents (natural or artificial) without a leader, for food. This is a classic example that has there is an advantage for an individual inspired the development of algorithms to sparrow to gain: survival. We can see how These agents have the capability of being simulate this behavior as we address later swarming is about animals organizing responsive towards each other, i.e. they on in this white paper. So, how do they do themselves to obtain the power of many. exchange data, either in direct contact or it? It starts with a common objective, being This property makes it interesting to see how indirect, e.g. through repositioning in a shared the search for food. To reach that goal, we can apply principles of swarm behavior in environment, ants start to move around, leaving behind the human domain. a pheromone trace to mark where they The agents have an individual task or goal, have been, serving as a memory. Initially 2.1.2 Swarm intelligence each of them wants to accomplish, which is they move randomly to cover the territory Swarm intelligence is derived from observing supportive to the goal of the group as a whole, and change direction when this does not the swarming group behavior of animals and lead soon enough to the discovery of food, trying to mimic this behavior, by capturing The ability to act (e.g. swim, walk, travel on applying the control principles (in a feedback this in rules, also called ‘heuristics’. In that the internet), to accomplish this goal or task mode). Once an ant has discovered food, sense it can be seen as a specific product and to conduct these actions in parallel from the ant will return and strengthen the of biomimicry which is looking at nature for each other, pheromones trace left behind, which attracts findings ways to solve human problems. the other ants to the food supply, again the Each agent applies a set of (simple) rules or control principles (in a feedforward mode). We define Swarm intelligence (SI) as “the heuristics to accomplish the goal or task in collective behavior of decentralized, self- an economic way (i.e. least energy consumed For an outside observer, these swarm organizing systems”, comprising of either in as little time as possible), activities are seen as self-organization, which humans or artificial artefacts [2.]. ‘Swarm emerges as a property of the behavior of intelligence’ was introduced by Gerardo To accomplish this goal or task, the agent the swarm as a whole and as key element Beni and Jing Wang in 1989, in the context has some form of memory (either internally of swarm intelligence. In the man-made of Cellular Robotic Systems (CEBOTs): a of externally) to evaluate its proximity from domain, we therefore consider how to apply self-organizing system composed of a large its objective / task end-state and adapt its swarm intelligence as a new organizational number of autonomous robotic units. behavior accordingly, paradigm, creating the conditions for self- organization to take place. SI systems consist typically of a population of The individual’s goal or task accomplishment agents, interacting locally with one another is to some extent dependent on some and with their environment. The agents each of the other agents and on their environment follow simple rules, and although there is no in a general sense, centralized control structure dictating how individual agents should behave, as a whole The agents have certain navigation skills they develop goal-oriented behavior. Local, and communication patterns to accomplish Self- Self- and to a certain degree random, interactions the goal, which includes a certain degree of healing configuration between such agents lead to the emergence randomness. of seemingly “intelligent” global behavior, which is more than the sum of the activities Each agent will apply its heuristics to reach of the individual agents. This intelligent group its goal, while exchanging information with Self- behavior is what we call ‘swarm intelligence’ other agents in its immediate proximity. optimization and is therefore considered as emergent This results in the desired behavior of the property of swarm behavior. group as a whole. SI is the resulting property of the group in relation to the group’s objective attainment.

Swarm intelligence 05 Obtaining swarm intelligence in man-made systems

3.1 From natural swarms to possibilities for man-made swarms to operate outcome in conjunction to the formulated artificial swarm intelligence within a hierarchical structure, where problem. The elementary rules cover both the higher level takes care of setting the feed-back and feed-forward mechanisms. We are interested in swarm behavior, not objective and constraints (like budget and SI-algorithms are selective in a sense that just because we study nature as such. From plan) for the group as a whole. The group successful behavior is amplified, and that a biomimetic perspective we ask ourselves then organizes itself to attain this objective, randomness is applied to obtain the variety what we can learn from animal swarms within the given constraints. of inputs necessary for being successful. to apply this in the human or man-made domain. Or, how to exploit the potential of In the next section we explore how SI Over the last decades, a large variety of the power of many and obtain the desired principles can be applied in the human algorithms has been developed (see for synergetic SI-effect? context and how SI-based systems can be instance [5.]). In the next section we explore used to resolve problems in a different way one of the most popular algorithms (i.e. PSO) In the man-made domain, we can adopt an than through a traditional approach. as an example of the approach followed by engineering approach to purposefully design the algorithmic school. systems in a way that they are going to exhibit 3.2 Schools of thought SI-features. In doing so, we are inspired by 3.3.2 Particle Swarm Optimization (PSO) swarm behavior in nature, but we need to take In exploring how man-made systems can algorithm as example into account some key differences between be designed to exhibit SI-features, three natural and man-made systems. schools of thought are found in literature and 3.3.2.1 Application practice, focusing each on SI from a specific PSO is applied to solve optimization problems, angle: i.e. find out the most appropriate solution in a set of candidate solutions. It is widely applied 1. Algorithmic school: Research in this to resolve a wide diversity of simulation, domain addresses the development and analytical and design problems (see [4.] application of mathematical algorithms to and [5.] for application examples). The goal resolve computational problems, inspired of the algorithm is to have all he particles by natural swarms. locate the optimum in a multi-dimensional 2. Organizational school: The subject of space of potential solutions. This is achieved First of all, the agents of a man-made swarm this school is to investigate how SI- by assigning initially random positions to all can be entirely artificial (like robots), people principles can be applied in the design of particles in the space and small initial random (in case of a group) or a combination of processes and structures in which people velocities. The algorithm is executed like a people and artificial devices (like mobile are the swarm’s agents, to overcome simulation, advancing the position of each phones). The types of interactions can problems encountered in classical particle in turn based on its velocity, the best therefore vary from exchange between organization models. known global position in the problem space robots, between people and between people 3. Automation school: This school studies and the best position known to a particle. and devices. how automated devices in the IT domain, like robots, computers, or programs can The PSO algorithm is applied for instance Another essential difference is in the apply SI-principles to improve execution in bioinformatics for which problems lay proximity feature of a swarm. Physical of specific tasks, often making use of the in the huge amount of data, and thereby proximity is a condition for animals to algorithms studied by the first school. computational complexity. As a result, a swarm. For humans or devices, just the satisfying approximation to the solution is often ability to mutually communicate (e.g. 3.3 Algorithmic school used instead of an unaffordable exact one. through the internet), is sufficient to enable self-organization, hence to obtain collective 3.3.1 Algorithms and their application 3.3.2.2 Principle behavior. An agent’s interactions are therefore The approach followed here is to develop How does PSO work? A subset of candidate not limited to its immediate neighbors only. algorithms, mimicking natural swarm solutions to an optimization problem is Principally, any agent can interact with any behavior. These algorithms are commonly defined as a swarm of particles which may other connected agent. Specific for human used in resolving certain optimization or flow through the possible solutions space agents, proximity also takes the shape of simulation problems through an iterative defining trajectories which are driven by being like-minded, or having a common approach, executed by the agents of the their own and neighbor’s best performances interest as a condition to swarm. swarm (or in this context often called evaluations. Over a number of iterations, “particles”). An SI-approach is applied either the particles of the swarm have their values Man-made systems consisting of artificial because the only way to resolve the problem adjusted closer to the member whose value devices are programmable. So, algorithms is iteratively, or the problem’s resolution time is closest to the target at any given moment. can be used to make these systems behave is shortened by making use of the processing in a way that they exhibit SI. Human behavior capabilities of the agents (instead of a by A particular’s behavior is determined by is not programmable in this way. Instead, we single central processor). a combination of two types of actions: need to provide the right sort of incentives (1) exploration, which is the behavioral that will stimulate people to behave in a The algorithm specifies a few elementary component that takes care of covering the swarm-like manner. This is what we will call rules, to be applied by every agent, which: search space and (2) exploitation, which takes ‘social contract’. (1) enables a decomposition and delegation care of utilizing the particles current position of the overall problem resolution task to the in terms of its target proximity. Exploration is Man-made systems are usually part agents, and (2) results in self-organization the random part of an element’s behavior. In of something bigger, often being an amongst the agents. In turn, this lead to a implementations of the algorithm usually a organization like a company. This creates the collective behavior that produces the desired variable is used so set the ratio between both

06 factors. The bigger the role of exploration organizations to the changing Semi-autonomous groups take on different is, the longer the process may take, but the environment, starting in the ’70’s, was to shapes in organizations, like task forces less likely it gets that not the best available handle the size and complexity issues. or incubators, but in general provide a solution in the solution space is reached. This was dealt with by concentrating degree of autonomy that enables and on core-business and strengths, while stimulates people to deliver a high-quality 3.4 Organizational school outsourcing or divesting non-core parts. performance in a volatile, unpredictable or However, the fundamental structure was unstructured environment. They are not a 3.4.1 How do human beings swarm? often not touched upon and for instance silver bullet for just any organizational issue. People swarm as well under certain rigid vertical chains of command remained Semi-autonomous groups provide another circumstances, but not driven by an sources of tardiness, conflicting interests organizational paradigm, that can be algorithm as explored in the previous and miscommunication. implemented in conjunction to the classical section. As addressed earlier, a key aspect model, there where this makes sense. of SI is the capability of self-organization, we Organizations which recognized this begun E.g. in those parts of an organization that therefore consider two basic forms of human to explore other models, in which the are directly exposed to the organization’s organization encountered in practice, human factor in terms of capabilities and environment, like R&D, Marketing and Sales. • Formally designed organizations, which motivational factors were more explicitly have been structured explicitly to reach a taken into account. In turn, this led to the 3.4.3 How informal organizations certain objective, like private companies, concept of semi-autonomous groups. can benefit from applying SI governmental institutions, or sports teams, We call these groups ‘semi-autonomous’ because their autonomy is limited to 3.4.3.1 Communities as a form of human , where • Informally evolving organizations the way in which they produce certain swarming objective and structure are not predefined, deliverables, which are used by the larger People join formal organizations on the basis but emerge as an outcome of the collective organization of which they are part. So, they of a negotiated contract (i.e. a committed behavior of the participants, who share a operate freely and regulate mutually their performance versus a salary). However, there common interest. activities within the boundaries of a given is still another type of contract for human assignment (see also [6.]). beings to organize themselves. This other type is based on people joining each other The self-organizational capabilities of the voluntary on an equal basis, because they group - as we have seen an essential property have similar personal objectives that cannot of SI - assure its perpetual orientation be achieved individually. A community is towards the common goal. Without central about people joining in this way on the basis management, the members of the team of what we call here a ‘social contract’, which divide the work amongst themselves, contains regulations on for instance code depending on progress and availability of of conduct or confidentiality, to protect the 3.4.2 How formal organizations can benefit team members. The obtained result is a interests of the participants. from SI form of memory (also called “stigmergy”) that is directly visible to everyone and As with animal swarms, the very first human 3.4.2.1 Limitations to the formal thereby supports the team members in communities may have originated because organization model deciding on and coordinating next steps. By of survival arguments (think about tribes). In the age of industrialization and expanding applying SI principles, the team becomes Nowadays we participate in a community markets, formal organizations with rigorously more resilient and efficient in attaining the because we want to share emotions divided tasks were a prerequisite to manage group’s required result than is the case in a or knowledge with others, or we want to growth, while making efficient use of scarce classical management structure. This applies achieve something together. goods, like machinery. Today, the environment in particular when dealing with a high degree of these companies has changed dramatically of uncertainty or turbulent environment that An audience in a soccer stadium forming in terms of its dynamics and complexity may affect the goal attainment. together a wave is an example of sharing and many markets have shifted from being emotions through which an individual person manufacturer- to buyer-driven. For semi-autonomous groups to be successful, feels to be part of something bigger. Being a number of conditions is to be satisfied (see with like-minded people creates feelings While Organizational structures in the past also [7.] and [8.]). Autonomy, congruence of safety, belonging and recognition. focused on stability, internal control and of objectives, open communication, multi- obtaining economies of scale, organizational skilled participants, team cohesion, shared Communities differ from the semi- structures today should accommodate for a accountability and short-cycled feedback autonomous groups we encountered state of permanent flux and focus at first and loops are ingredients of a successful team. in the previous section. A community foremost on serving their environment. In can be fully autonomous in a sense that it this situation size and structure have become A good example of such a semi-autonomous can define its own objective, whereas this is an obstacle as they created inertia in dealing group in the IT domain is a DevOps team. In a given for a semi-autonomous group being with change. So, how have organizations essence, a DevOps team is a self-organizing part of a larger formal organization responded to this?. group in which everyone manages his/her to which it needs to contribute. Consequently, own activities in a way that they are aligned the social contract for a community 3.4.2.2 Semi-autonomous groups to the activities of the others in the team can be a rather lose one and individuals can for agility where needed and that they contribute to producing the freely leave or join a community as they Initial response of western industrial required deliverable. please.

1The concept of semi-autonomous groups is not new. Experiments with this concept began in the 1960’s, done by the British Tavistock Institute, building upon earlier research at the Hawthorne Works (in Chicago) between 1924-1932 into the relationship between productivity and motivational factors.

Swarm intelligence 07 3.4.3.2 Swarming in the digital world In the last two examples a digital community a basis for trust. We expect blockchain As in animal swarms, the early human is used to establish connections in the platforms to provide the additional support communities required individuals to gather physical world. This is done by creating a layer for using the internet in a trusted way. physically to obtain the required level of virtual proximity first, which can then be This enables people, being remote and self-organization. Individuals being in direct followed by the physical proximity. The digital strangers to each other, to cooperate and proximity of each other was a prerequisite community is used in this way as a low conduct transactions amongst themselves for having swam-type of direct interaction. threshold facility enabling individuals to judge as they would do when being physically The coming of the personal computer, the whether they want to make the additional together. Neither is there the need for internet and smartphones have changed this and more substantial effort of joining a trusted middleman or central regulating dramatically. Being together physically is no physically. All mentioned examples have agent, as found in the traditional models, longer necessary to connect. very clear objectives and attaining these to compensate for some of the weaknesses objectives depends on sufficient participation inherent to the digital world at present. As long as someone in the “digital world” is of like-minded individuals. Blockchain will enable a community to reachable via a digital address, anyone can develop the closeness (or virtual proximity) participate in any community, regardless of 3.4.3.3 What are prerequisites for successful that is needed for people to swarm and location. communities in the digital world? bundle their power. For digital communities to exhibit self- But what kind of communities in this organization properties, the same key 3.4.4 What future organizations Digital world could be qualified truly as ‘SI- principles apply as for semi-autonomous will look like communities’? groups: clear overall objective, close As the digital world matures, we can expect alignment with personal objectives of that communities professionalize as well, What is called a ‘community’ in the digital participants and a homogenous composition for instance through adopting SI-principles. world can in many cases hardly be called a of the group. In particular in digital They will institutionalize, become more community in the sense as we just described. communities, when people are remote, successful and will more clearly differentiate Often internet communities, facilitated by the trust is an important condition for these from non-SI type of social gatherings on familiar social media companies, are places communities to be successful. Trust takes the internet. These new communities will where people connect to, just to ventilate time to develop and traditionally requires also develop their own economic models opinions and given some proof of their people to meet face-to-face. Trust stimulates and incentivization mechanisms, leading existence. Mechanisms of self-organization someone to invest in a relationship and to to new types of social contracts. This to and clear objectives leading to convergent take risks in exploring new avenues. The provide the necessary stability and trust base behavior are often absent. higher the trust amongst individuals, the for participants to work together on more higher the ambition level can be set for the ambitious objectives, than merely chatting However, there are examples of successful community as a whole. Prerequisite is that and exchanging pictures. As the integration digital communities in which we can people equally contribute to this and trust of the digital world into the physical world recognize SI-principles: each other to do so. progresses, these communities could • Buying groups, consumers connecting become the foundation for the companies with the objective to obtain best possible When people work physically together of the future. We might see next how some prices and conditions for certain products and in a formal organization, measures are of these communities are going to swarm or services. These groups may exist only available to obtain the desired behavior, themselves, that is grouping with other temporary, initiated by someone who has a e.g. through the formal contract and social communities, when synergetic advantages direct need to acquire a product or service. control. However, when people are remote can be obtained. By joining the group, one can get a better only connecting through the internet without price than operating individually. knowing each other face-to-face, misbehavior Organizations adopting (selectively) SI is more likely to occur and trust much more principles, will be better fit for the future • Knowledge groups, the best example is difficult to develop. We see in the digital world than organizations that stick to more Wikipedia that provides the platform for how communities often have a temporary conventional models solely. The need for authors to group around a specific topic nature and fail to reach their objectives, an organization to be more responsiveness and contribute to its documentation. as participating individuals fail to develop to an increasingly uncertain and volatile Jointly the authors take care of expanding sufficient trust in the other participants, and environment, the shift towards knowledge- and improving the quality of the then decide not to contribute as needed. based economies, almost everyone documentation, relying on a stigmergy and everything being connected in kind of memory. Individuals and companies can use today’s the digital world, growing flexibility and • Shared economy, people living in the digital world to create organizational commoditization of technology, are just a same neighborhood make their property settings, for joining forces to reach individual few trends that put classical organization (e.g. power tools, cars) available for lending objectives through a group performance. models under pressure. Combining these amongst the participants without any Apart from the factors mentioned in this trends, we can expect to see increasingly charge. People can only borrow when they section, what is needed to make this less large-scale companies in the future (see make some of their properties available for happen, is a digital platform for people and also [9.]). Specialized companies will shape lending to the others. companies to exchange transactions in a safe up by bringing workers virtually together • Meeting groups, people sharing an interest and trusted way. These transactions embody to work for a certain period together on and living in the same village or region new types of social contracts for working particular objectives or tasks in a swarm-like use the internet to group around topics for together and support self-organization as manner. Companies like these will need to which they want to organize face-to-face described earlier. liaise with other companies, each covering gatherings. This can be social meetings a particular part of a business chain. Like or working groups for instance to share Blockchain can provide the concept for such with the workers, these liaisons may be of views. However, this type of swarming may a platform. The way blockchain transactions temporary nature, thereby forming swarms also take a more destructive shape, e.g. are processed amongst participants provides (or “ecosystems”) of companies. hooligans gathering for a riot.

08 3.5 Automation school • Multiple inputs: collecting data about swarm robotics systems that need to the working environment or surrounding reach an agreement making sure that all An autonomous system is defined here as environment of a swarm robotics system is individuals continuously share an identical a man-made system that can independently an everlasting job. view of the world. compose and select among alternative • Resilience: a swarm robotics system needs • Knowledge acquisition and transfer: courses of actions to accomplish goals based to work continuously towards the objective any individual joining a swarm robotics on its knowledge and understanding of the of the system’s task without significantly system will be promptly get trained and world, itself, and the dynamic local context. degrading performance, in case of synchronized with the rest of the swarm by It must be able to respond to situations that individuals fail for some reason, to deliver downloading the history of all agreements are not pre-programmed or anticipated prior the right outputs. and all related transactions, and knowledge to its deployment. already discovered and stored in the • Scalability: individuals are substitutable blockchain. and can therefore join or quit a task at any There are numerous advantages to using time without interrupting the whole swarm • Behavior differentiation: swarm robotics swarm principles in designing autonomous robotics system that can self-adapt to the systems deployed to achieve different tasks systems, including: change in population through implicit task will need to handle different behaviors • Parallelism: tasks can be decomposed re-allocating schemes without any external according to these tasks purposes. and performed in parallel rather than operation involvement. It can be handled by the capability of sequentially, blockchain technology to link together 3.5.2 Usage of swarm robotics systems several blockchains that will allow a swarm • Efficiency: tasks can be decomposed in Various types of applications of swarm such a way, that different agents of the robotics system to act differently as per the robotics systems have been found valuable particular blockchain being used. system can cooperate together to obtain by academics, industry and military over de • New business models: one of the most a better end-result than individuals might recent years, as demonstrated in real-life, like: achieve, common usages of blockchain technology • Underwater habitats: The EU-funded • Fault tolerance: the failure of a single agent is cryptocurrencies. It is an ideal Application Collective Cognitive Robotics 2011 project Programming Interface for economic within a group does not necessarily imply [12.] aimed at developing a swarm of that a given task cannot be accomplished. applications that will allow swarm robotics autonomous underwater vehicles that systems to directly take part in an are able to interact with each other and economy. Potential applications for swarm automation which can balance tasks such as ecological are many and some are described in the next monitoring, searching, maintaining, 3.5.4 Which future for automation school? sections. They include tasks that demand exploring and harvesting resources Swarm intelligent autonomous systems miniaturization like distributed sensing tasks in underwater habitats. are one of the most fascinating research in micromachinery or the human body, and application areas of recent decades. • Space exploration: NASA has investigated and activities that demand cheap designs, It perfectly fit the emergence of the swarms for future space exploration for instance mining or agricultural foraging. networked connection of people, process, missions [14.]. Given that many of the It is also considered that one of the most data and things that is exploding, pushed spacecraft could collide with one another promising uses of swarm automation is in by either Business-to-Consumer (B2C) or or with asteroids and become lost, disaster rescue missions (see [11.]). Another Business-to-Business (B2B) business models, multiple-spacecraft missions offer greater large set of applications is autonomous for which current vertical architectures likelihood of survival and flexibility than surveillance and reconnaissance. U.S. will not work anymore. As computing single-spacecraft missions. Additionally, the Naval forces have also tested a swarm of capabilities in the cloud and at the edge self-directed swarm will exhibit intelligence, autonomous boats that can steer and take become increasingly intertwined, we will which is critical since round-trip delays in offensive actions by themselves (see [11.] see the emergence of large computing communication from Earth can stretch to and [13.]). continuums. Much as we see swarms in tens of minutes. nature acting as intelligent collectives, the 3.5.1 Value of swarm robotics systems • Pollutant absorbing: researchers of individual computing capacity of the devices A robot is usually inspired by human Massachusetts Institute of Technology at the edge will be complemented and behavior. As it is hard to simulate the have developed a fleet of low-cost oil supplemented by their dynamic cooperation human interactions, such robots are often absorbing robots called ‘Seaswarm’ for with other objects in swarm communities sophisticated, complex to design and ocean-skimming and oil removal [15.]. acting as an intelligent whole. consequently expensive. Comparatively, The system provides an autonomous and Swarm Robotics is inspired by social animals low-cost solution for ocean environment This key trend of swarm of connected for which the mimicking of animal groups is protection. computing devices will only get successful simpler to simulate than human behaviors. is the Artificial Intelligence capabilities of As the individuals in a swarm are reasonably 3.5.3 Improve trust and traceability with distributed systems are evolving accordingly. simple, small and low cost, it gives swarm blockchain We will need by example to evaluate robotics a bright future in dealing with large The properties of swarm robotics systems learnings from individual participants and scale operations or environments, offering can be greatly leveraged by the blockchain enforce and spread out the ones that are novel approaches: technology [10.], [18.]: considered as the most promising. • Goal seeking: a swarm robotics system’s • Security: individuals of a swarm robotics behavior is driven by the objectives of the system need to safely detect and The future of swarm intelligent autonomous tasks they are assigned to. trust their counterparts. A blockchain systems is definitely paving the future of the B2C and B2B business models. • Control loops: a swarm robotics system as technology can fulfill these objectives, a whole has to stay focused to achieve proving the identity, guarantying reliable the task it has been assigned to, whatever peer-to-peer communications, making the conditions and the events encountered transactions using unsafe and shared by any individual of the swarm. channels, overcoming potential threats, vulnerabilities, and attacks. • Memory: a swarm robotics system relies on communication between adjacent • Distributed decision making: blockchain individuals that maintains local knowledge technology allows for the possibility of shared between neighbors. creating distributed voting systems for

Swarm intelligence 09 Is swarm intelligence going to be a game changer?

4.1 It’s time to engage with nature 4.2 Will your company be a self-organization and they have to perform disruptor or will it be disrupted? as swarm-organizations: team individuals Did you know the Eiffel tower, 324 meters tall are appointed to a higher occupational with a ground square of 125 meters’ width, While many companies are struggling to competency and have freedom of action. It was inspired by your femur, the lightest attract, and retain top millennials (Gen Y), requires greater flexibility and self-motivation, and strongest bone of the human body? they are starting to see another tsunami, working as internal entrepreneurs within their In addition, if you imagine a cylinder that potentially just as disruptive: Gen Z. own company, which is a key driver for Gen Z completely wraps around the Eiffel tower, the motivation, involvement and satisfaction. air in this tube outweighs all the iron in the But what’s Gen Z? Gen Z, people born from tower. the mid-1990s to the early 2000s is the Although the gig economy is not yet first generation to be raised in the era of widespread, it is gaining traction of the smartphones. They have been exposed to workforce, specially to gen Z with their main Biomimicry: turn to nature for innovative design solutions an unprecedented amount of technology differentiator of attitudes and behavior and in their upbringing and do not remember for which flexibility is a key driver for seeking a time before social media. Gen Z valuable a new employment experience [16.], [17.]. characteristics are their acceptance of new Working in multiple roles such as part-time, ideas and a different conception of freedom freelancing or in temporary assignments from the previous generations [16.]. rather than a traditional full-time position will become increasingly common. Rather than Gen Z is also driven by YOLO: “You Only Live investing in expensive, inflexible long-term Once”. In other words: be spontaneous and employee contracts, companies might transparent, always communicate openly and temporally bring in individuals with specific honestly, and live life to the fullest. They don’t skillsets as needed, or grow and shrink their want to waste time with things they don’t workforce as demand, resources, profitability enjoy. Corporates need to understand how etc. The gig economy could appear to yield to embrace the YOLO entrepreneurial spirit, benefit for both workers and companies. developing new models of lifelong learning, a It surrounds us, the secret to survival: there combination of hands-on experience, training Blockchain’s decentralized model enables is no better designer than nature. Through and mentoring. secure peer-to-peer transactions without billions of years of evolution, life’s products, so the need for intermediaries or centralized to speak, have been extensively prototyped, Lastly, Gen Z is as the heart of the digital authority. It also makes it possible to execute market tested, upgraded and refined. revolution and feels definitely comfortable self-executing contracts that are called. with the digital world. Waves of disruptions If and when the pre-defined rules are met, Back to the Eiffel tower, you don’t have to based on digital revolution they are born with the agreement is automatically enforced. go to Paris to see it. Just look down at your are increasingly revolutionizing traditional It is the simplest form of decentralized feet. In 2017, scientists at Georgia Tech industries. Any company has now to automation [18.]. For the gig economy, this investigated how ants built themselves into manage fundamental changes in the way combination of blockchain transactions towers to escape confinement or danger [19.]. its deals with customers, the way it builds and smart contracts makes it possible to They discovered that the ants climb upwards, its brands, the way it processes, distributes, create freelancing and resource-sharing on top of each other, until they find an empty markets, analyzes information, and the way platforms where employers can find and hire spot. Then they stop. The next one does customers access its products and services. employees and compensate them without the same, and so on. As the tower grew, the It becomes more about dynamics and agility the need for a broker. Payments are made ants in the middle start sinking and climbing to keep ahead of the competition rafter than immediately in cryptocurrency without any out from the bottom. They then restart their efficiency and stability which are the essence delays. journey to the top as they continuously of traditionally hierarchical organizations. rebuilt the tower. It inspired the researchers This hyper-competition is driven by multi- We could even consider smart contracts who believe that this construction method dimensional innovation capabilities, at least associating freelancers and investment funds could be useful in programming swarm two different types of innovations being competing against classical companies on intelligent robots for building purposes. interlaced and promoted simultaneously: a deal-per-deal basis. In such environments, products and services on one hand and corporates would not face anymore a Since the beginning of the humanity, business models one the other hand. situation where the gig economy brings our story has been driven to gain our them advantages, but competitors with the independence against nature. Biomimicry This new multi-dimensional competition for best experts, great flexibility and agility, and becomes more than just a new way of innovation as well the Gen Z expectations backed on credible financial capabilities. It looking at nature. This sustainable design are strong organizational challenges for could even shake the long-term future of approach becomes a race and a rescue. companies. It is not only definitely required corporates. to build interdisciplinary teams beyond Are you ready to create man-made swarm the traditional departments and already It’s time to ask you: will your company be intelligence based on the biomimicry established hierarchies, but also teams that a disrupter or will it be disrupted? revolution that introduces an era based not are capable of quickly acting and reacting on what we can extract from nature, but on to unforeseen changes. Efficiency of such what we can learn from her? teams relies on peer-to-peer connections and

10 4.3 Take the right decisions Gen Z driven by YOLO: “You Only Live Once” faster than competition • 72% want to start a business someday • 61% want to be an entrepreneur and not an employee Gen Z has grown up in the middle of an era of developed information technology, constantly engaging in online exchanging Distributed of information and conversation among Autonomous Projects its peers. They value career growth and Distributed Autonomous Cheaper without development above anything else when it Enterprise a firm comes to employability. They are in a world Business WEB Cheaper outside where environments are complex, changing Focus on what the boundaries • Best experts all the time where the global optimum you do best of the firm • Great flexibility Extended possibilities and value may change with time. Enterprise • Agility They want systems that help them to take a Industrial Age Cheaper inside • Backed by credible decision on the fly. Corporation the boundaries of the firm financial Mass production capabilities In the same way, with digitalization and disruptive technologies, businesses are Shake companies future challenged all the time and they need systems which can help them decide on the direction to take in these complex systems. feeling comfortable with our company, require are driven by algorithmic capabilities (like from us agility and instantaneous appropriate PSO). Organizational decision making in The expectations from the systems are that actions and reactions. this way becomes a perpetual rather than a they should enable them to derive insights periodic process, enabling an organization to and help the business identify directions in Collective Humans coming together as a continuously adapt its direction and actual a highly non-linear, correlated parameters swarm will even form a hive mind – a ‘brain state of affairs in response to a highly volatile environment. With traditional mathematical of brains’. With machines augmenting the and uncertain environment. Adopting this modeling, finding an accurate estimate for all ‘brains of brains’ governed by above trends, capability can become a matter of survival the parameters is very time consuming and like AI, these combine the knowledge, for an organization, as we have seen flocking almost impossible for such complex models. wisdom, insights, and intuitions of real is for a swarm of sparrows. people and in real time, enabling large SI algorithm’s, like PSO we discussed earlier, swarms of networked humans to quickly Let swarm intelligence and PSO like can be deployed to track this dynamic converge upon optimized decisions, algorithm promises to make our life and shifting optimum. It will help to work predictions, solutions, and evaluations that business easier. together in synchrony, forging a unified dynamic system that can help them make decisions by exploring a decision-space and converging on a preferred solution. Volatility Uncertainty Complexity Ambiguity

These systems will allow the user to perceive Being a large organization Knowledge, wisdom, insights, • On time decisions and react to the changing system in real-time is no longer an advantage and intuitions combined in real time • Continuous adoption integrate noisy evidence, weigh competing Speed is essential to survive Large swarms of networked • Multi alternatives - simplified alternatives and converge on a reasonable heuristics optimum quite faster than any other approach humans quickly converge upon optimized predictions and at a competitive computing cost. and decisions driven by swarm algorithms (like PSO) Our world is moving extremely fast and decisions to conduct businesses are Not the best, but fastest good enough ones stay ahead of competition more and more difficult to take in such an environment in which we have to weight tons of considerations. We would wish some time to make sure we take the right ones. However, both staying ahead of the competition and making the Gen Z who is paving our future

Swarm intelligence 11 4.4 Create top class value from an ultra-connected world • Managed Billions of infrastructures Networks Autonomous & intelligent Waves of disruptions based on digital • Hybrid cloud thousands Millions of Objects & Machines revolution have increasingly revolutionized • IT security of IoT/Edge Systems 75% of enterprise data Hybrid Clouds traditional industries in the B2C sector, • Cloud security created and processed starting in the media industry and trade. Out-of-Cloud in 2022 The industrial revolution in the B2B world didn’t however take-off so early and has Today Tomorow only recently started with the emergence of intelligent machines essentially defining Infrastructure of infrastructures Traditional IT ‘Industry 4.0’. It is estimated that smart Exponential critical applications, intelligent computing and security needs machines and robots will perform 45% of all manufacturing tasks by 2025, compared to about 10% in 2016 [20.]. It requires us to consider the perspective of in conventional parallel processing. With IoT This is also driven by the terrific growth self- and swarm-operating machines with providing information about assets and with of data. By 2020, there will be more than capabilities of data analysis and context- objects becoming increasingly self-aware, 16 zettabytes of useful data (16 trillion GB), specific decision-making that are beyond the sharing platform of the future could have reflecting a growth of 236% per year from the level of human performance. The capabilities of permanent availability for use 2013 to 2020. The Internet of Things (IoT), consequences are manifolds and we have in real-time and efficient decision making. meaning the things that are networked to carefully anticipate benefits and risks, Enhanced tagging and tracking capabilities together and produce data, is driving this including ethic and societal impacts. with swarm intelligence, present enormous data explosion. IoT applications will yield a economic opportunities to prolong the life of vast quantity of data that has to be analyzed, This hyper-connected and distributed world goods, plug leaks and make use of materials and can be analyzed with recent advances in is resulting into data and more importantly previously considered to be waste that computing power and connectivity [21.]. information and next knowledge becoming contribute to promote a circular economy. new gold mines. These mines are exploited Nevertheless, a complete and efficient using new avenues of digital platforms. Most of the 16 zettabytes of useful data we industrial product or process in its whole A digital platform can be thought of as a mentioned will be generated and kept at life cycle has be accompanied by a virtual set of technical building blocks that act the edge, moving away from centralized representation, often called “digital twin” as a foundation upon which an array of data storage concepts. Challenges will that allows design optimization, process firms (a business ecosystem), can develop be around the control of these data, like control, lifecycle management, predictive complementary products, technologies or sensitivity to intellectual property, access maintenance, risk analysis and many other services. Data is used there to bring context to and privacy of data. Digital giants are features [22.]. This digital twin bridges the and insight, identifying potential business attempting to monopolize its monetization gap between the physical and digital worlds, moments that are subsequently exploited. through controlling ownership rights of relying on the connections at the interface Now, applying this to swarm intelligence, even personal data assets by various between fields of expertise, domains and these platforms will become more intelligent means, including some of dubious legality, businesses, and across the complete lifecycle and will adopt creating differentiation morality or legitimacy. This is creating new of products and systems, improving speed to through improving understanding and agenda items for the public and political market and efficiency. situational awareness when preparing the debate. Enforcing open API access to certain next generation of its products and services. data types (e.g. PSD2 in Europe and UPI It takes place in the digital revolution under in India), and EU laws relating to personal way for which the agents are proliferating, These data platforms with disruptive digital data privacy (GDPR) are all moves to avoid e.g. Edge Computing, Big Data, Industrial technologies will enable new business such anti-competitive behavior and protect Data Platforms [23.], High Performance models. SI characteristics of resiliency, citizens. Consequently, there will be some Computing, Blockchain [24.]. These agents decentralized, self-repairing and scalability organizations who take a nervous approach are enabling the transformation of services without experiencing complexity problems, to this and secure all their data, treating and manufacturing and reshaping entire present immense opportunities to resolve them as personal data, and inevitably sectors of the economy. problems or complete tasks. This is done blocking new opportunities. through cooperation between local elements This shift is starting from the current vertical all working on the same overall objective, Is your company on track to create top value architecture promoting the cloud as owning all rather than having tasks distributed from your information assets and rush ahead the intelligence, collecting data from sensors in a compartmentalized way as we see of the competition? and sending commands to the actuators,

creating a cognitive loop. Then, we are moving Data to the world of the edge computing with smart machines autonomously running on Neural interfaces, genomics… the field functions that delegated from the Digital augments humans 2030 cloud, leveraging decisions that are taken as IoT, AI, real-time, cloud… close as possible to the point of action. Last, Digital interconnects things Web, Mobile, … 2010 the fog computing era is coming with the We are here Data drives life Computers, ERP, … Digital connects people intelligence spread everywhere. Transversal Digital manages processes 1990 Data drives business 1970 cooperation is definitely supplanting vertical Digital enterprise integration, hierarchy with systems running altogether Data drives customer Businesses model evolution, full processes as a swarm intelligence, having Data drives processes experience Ecosystems transformation … the ability to operate out of the control of the cloud. IoT drives the 3rd digital wave Time

12 Conclusion

Is swarm intelligence going to be a game changer? No doubt, the answer is YES.

Swarm intelligence involves a different way of looking at things, whatever it is exercised by human or machines leveraging the capability of a group to be smarter when thinking and acting together than individuals would be on their own.

In our complex world and fast-paced, Conclusion - This is just the beginning constantly evolving environment, it is more of the data revolution 1 than ever time to act as a person Tomorrow’s opportunity: from big to Yottabyte or a company smarter than others. Let’s be gigantic data around swarm intelligence minded to better: answer questions, make predictions, optimize 2030 opinions, generate forecasts, reach decisions, 1 Yotta=1000 Zetta and so on. 175 So, no matter which way you look at this, Zettabytes in a world in which almost all people and in 2025 devices are connected, we demonstrated We are 50 Using 20-25% of worldwide energy production we should start to think about how to exploit here Zettabytes this huge potential and develop the SI-based Zettabytes 33 in 2020 80% enterprise data created out of cloud, 30% IOT real time, EDGE everywhere digital solutions and human being. We now in 2013 Zettabytes in 2018 50 billions connected objects and machines. Digital industries, workplaces, cities,… have the basic technology to make it happen 4 and there is a natural fit between SI and the Has doubled every two years Everywhere Cybersecurity zeitgeist. It is this coincidence of trends in Sources Gartner, IDC, Statista, Atos 1 ZETTABYTE = 1,000,000,000,000,000,000,000 bytes society, business, economy and technology that we belief is likely to set the scene for intelligence is going to be a game changer. SI to break-through in organizations and in Each of them perfectly fit one or several digital solutions and services, of which we of our SI predictions: mentioned some in this document, and to It’s time to engage with nature become mainstream. Will your company be a disruptor or will it be disrupted? Swarm intelligence is already at stake in Take the right decisions faster most of the influential technologies and than competition associated business opportunities and Create top class value from an challenges Atos put forward in its vision ultra-connected world Journey 2022 in which we expect to see a post-digital world that has a greater focus on understanding and responding to the more human, real-world implications of Ready to reap the digital and business technologies. All these rewards of swarm challenges deeply discussed in Journey 2022 are fully connected to our statement: swarm intelligence promises?

Swarm intelligence 13 Authors

Eric Monchalin Head of Machine Intelligence [email protected]

Purshottam Purswani Chief Architect [email protected]

Do van Rijn Bid Executive [email protected]

14 References

[1.] Swam behavior, December 18, 2018, definition on Wikipedia: [14.] Rouff, C., (2007). Intelligence in Future NASA Swarm-based Missions. https://en.wikipedia.org/wiki/Swarm_behaviour Available at: AAAI, Inc., https://www.aaai.org/Papers/Symposia/ [2.] Swarm intelligence, December 18, 2018, definition on Wikipedia: Fall/2007/FS-07-06/FS07-06-021.pdf https://en.wikipedia.org/wiki/Swarm_intelligence [15.] Quick, D., (27 August 2010). MIT researchers develop autonomous oil- [3.] REVEALED: The REAL reason birds flock in huge numbers. absorbing robot to clean up oil spills. Sunday Express. At: https://www.express.co.uk/news/nature/755251/ At: New Atlas, https://newatlas.com/seaswarm-autonomous-oil- Revealed-real-reason-birds-flock-huge-numbers absorbing-robot/16153/ [4.] Poli, R. (2008). Analysis of the Publications on the Applications [16.] Fries, K., (25 September 2017). 5 Reasons Why Millennial Leaders Need of Particle Swarm Optimization, Journal of Artificial Evolution and Performance Feedback. Applications, Volume 2008. doi:10.1155/2008/685175. At: Forbes, https://www.forbes.com/sites/kimberlyfries/2017/09/25/5- reasons-why-millennial-leaders-need-performance-feedback/ [5.] Zhang, Y., Wang S., & Ji, G. (2015). A Comprehensive Survey on Particle Swarm Optimization Algorithm and Its Applications, Mathematical [17.] PERSOLKELLY (2018). 2018 APAC Workforce Insights, Gig Economy: Problems in Engineering, vol. 2015. http://dx.doi.org/10.1155/2015/931256 How Free Agents Are Redefining Work. Available at: https://www.persolkelly.com/media/persolkelly/ [6.] Delarue, A., Stijn, G., & Van Hootegem, G. (2003). Productivity client/2018%20APAC%20Workforce%20Insights%20-%20Q1/ outcomes of teamwork as an effect of team structure. Working paper, PERSOLKELLY%202018%20APAC%20Workforce%20Insights%20 Steunpunt OOI, Catholic University of Leuven. Available at: -%20Gig%20Economy%20How%20Free%20Agents%20Are%20 http://www.ondernemerschap.be/documents/pdf/wp_productivity_ Redefining%20Work.pdf outcomes_of_teamwork.pdf [18.] BlockchainHub. Smart Contracts. https://blockchainhub.net/smart- [7.] Vašková, R. (11 February 2007). Teamwork and high performance work contracts/ organization, Eurofund. Available at: https://www.eurofound.europa.eu/publications/article/2007/teamwork- [19.] Phonekeo, S., Mlot, N., Monaenkova, D., Hu, D., & Tovey, G. (12 July and-high-performance-work-organisation 2017). Fire antsperpetually rebuild sinking towers. Royal Society Open Science. Available at: http://rsos.royalsocietypublishing.org/ [8.] Landy, F. J., & Conte, J.M. (2012). Work in the 21st Century: An content/4/7/170475 Introduction to Industrial and Organization Psychology. 4th edition. Wiley-Blackwell. [20.]Bank of America Merrill Lynch (December 2015). Robot Revolution: Global Robot an AI Primer. Primer. Thematic Investing. Available [9.] Intuit Inc, (October 2010). Intuit 2020 report. Available at: through: https://www.coursehero.com/file/p64o163/Bank-of-America- http://http-download.intuit.com/http.intuit/CMO/intuit/ Merrill-Lynch-Robot-Revolution-Global-Robot-AI-Primer-Primer/ futureofsmallbusiness/intuit_2020_report.pdf [21.] Turner, V., Gant, J.H., Reisel, D., & Minton, S. (April 2014). The digital [10.] Ferrer, E.C., (June 2017) The blockchain: a new framework for robotic universe of opportunities: rich data and the increasing value of the swarm systems. Cornell University. Internet of Things, Report from IDC for EMC. Available at: Available at: https://arxiv.org/pdf/1608.00695.pdf https://www.emc.com/leadership/digital-universe/2014iview/index.htm [11.] Grayson, S., (8 May 2014). Search & Rescue using Multi-Robot [22.] European Service Network of Mathematics for Industry and Systems. Available at:https://www.maths.tcd.ie/~graysons/documents/ Innovation (23 April 2018). Modelling, Simulation & Optimization COMP47130_SurveyPaper.pdf in a Data rich environment. Available at: [12.] Robohub: Tracking the development of the world’s largest http://www.eu-maths-in.eu/EUMATHSIN/2018-new-modeling- autonomous underwater swarm. simulation-and-optimization-in-a-data-rich-environment/ At: http://robohub.org/cocoro-tracking-the-development-of-the-worlds- [23.] Beetz, K., Esteban, J., Hahn, T., Hall, J., Lehmann-Brauns, S., & Tardieu, H., largest-autonomous-underwater-swarm/ The Rise of Industrial Data Platforms. [13.] El Zoghby, N., Loscri, V., Natalizio, E., & Cherfaoui, V. (2014). Robot [24.] Hall, J., Kozakiewicz, N., Rosauer, B., & Tardieu, H., Trust and Traceability Cooperation and Swarm Intelligence, Wireless Sensor and Robot Technologies for Industrial Data and IoT Networks: From Topology Control to Communication Aspects, p.168- 201. World Scientific Publishing Company. Available at: https://hal.archives-ouvertes.fr/hal-00917542/document

Swarm intelligence 15 CT_J1917_190613_RY_WP_SWARMINTE About Atos Paper White of theAtos group. June2019©Atos. Confidentialinformation Atos, theAtos logo, Atos Syntel, andUnifyare registered trademarks Let’s adiscussiontogether start atos.net/career atos.net Find outmore aboutus space. technology information in the confidently and sustainably develop and at large to live, work societies of members and collaborators, and employees customers, its group the enables Across world, the excellence. technological and tocontribute scientific to research that approaches pluralistic and as well as multicultural education knowledge, of development the support services and expertise Its space. technology information future the of Atos of the is to design purpose help The stock index. Paris CAC40 the listed on Europaea), (Societas Atos, Atosbrands Syntel, Unify. and Atos is aSE the under operates and Games & Paralympic Olympic for the TechnologyInformation Partner group Worldwide is The the solutions. Workplace Digital and Applications Business Data, Big provides Orchestrated end-to-end Hybrid Cloud, Group the Computing, High-Performance and Cybersecurity in Cloud, one number European revenue annual and over of billion. €11 countries in 73 employees overwith 110,000 transformation in digital Atos leader is aglobal distributed norquoted without priorwritten approval from Atos. may notbereproduced,or any ofit, part copied, circulated and/or owned by Atos,to beusedby therecipient only. This document,