The Biological Principles of Swarm Intelligence

1. The Biological Principles of Swarm Intelligence Presenter Name : Pradeep B Presentation Date : 02-25-2014

2. Overview • Introduction • The underlying mechanisms of complex collective behaviors • Categorizing the collective behaviors of social insects • Conclusions

3. What is Swarm Intelligence ? • Swarm intelligenceis the discipline that deals with natural and artificial systems composed of many individuals that coordinate using decentralized control and self organization. • In particular, the discipline focuses on the collective behaviors that result from the local interactions of the individuals with each other and with their environment.

4. Examples of Swarm Intelligence • Colonies of ants • Schools of fish • Flocks of birds

5. A single ant or bee isn’t smart ,but their colonies are The study of swarm intelligence is providing insights that can help humans manage complex systems form truck routing to military robots.

6. Large Scale Networks

7. Large Scale Structures • If termites were the same size as human beings, their nests would be twice as high as the Empire State building, seen above.

8. Complex Structures • Because of the perfectly functioning air-conditioning system within the nest, its internal temperature varies by only half a degree throughout the year. The temperature of the air in the horizontal channels is lowered, and the air is then transported to a deep cellar; rising hot air takes its place.

9. The Underlying Mechanism of complex collective behavior • The anthropomorphic hypothesis • The complexity of behaviors and patterns observed at the colony level should be a direct consequence of the individual ability • To centralize information about the environmental conditions • To build on internal representation of these conditions • To choose the appropriate actions to perform

10. A hierarchical and centralized organization? Information coming from the colony is gathered and monitored by the queen which then controls and supervises workers activities

11. The society has no supervisor • Individual insects do not have a mental blueprint of the architectures they build or a global representation of the state of the colony • Their cognitive system is not enough powerful for a single individual to assess a global situation, centralize all the information coming from its colony and then control the tasks to be done by the other workers

12. Social insects colonies are distributed information processing systems Local information The rules specifying the interactions among insects are executed on the basis of purely local information, without any knowledge of the global pattern A limited set of instructions Each insect is following a small set of simple behavioural rules (≈ 20 elementary behaviours in ants) Emergent cooperation Complex collective behaviours emerge from interactions among individuals that exhibit simple behaviours. Social insects can solve problems in a very flexible and robustway

13. Pierre Paul Grasse (1895 – 1985) The Stigmergy “An insect does not control his own work but its on going activity is guided by the product of its work”

14. Stigmergy : Invisible writing • Stigmergy occurs when insects actions are determined or influenced by the consequences of another insects previous activities • This is a form of indirect communication that makes possible the coordination and regulation of insects activities. • This process leads to on (almost) perfect coordination of the collective work and gives us the impression that a colony as a whole is following a pre defined plan.

15. Next Building in social wasps : A stigmergic behavior • Wasps nests are built with wood pulp and plant fibers • Individual construction behavior can be studies in great details such as the wasps decisions to build a new cell in particular locations on the comb • The nest sometimes controls the organization of building activities • To decide where to build a new cell wasps make use of the information provided by the local arrangement of cells on the comb

16. The control of nest building in wasps : potential building sites on a comb • New cells are not added randomly to the existing structure • Wasps have a greater probability to add new cells to corner area (3 or 4 adjacent walls ) than to initiate a new row by adding a cells on the side of an existing row( 2 adjacent walls)

17. Modeling nest building : Behavior of the virtual wasps • Wasps are modeled by asynchronous automate with a stimulus response behavior • Virtual wasps move randomly in a 3D discrete hexagonal lattice • Virtual wasps only have a local perception of their environment ( the first 26 neighboring cells close the cell occupied by the wasp) • Do not have any representation of the global architecture they build

18. Local construction rules followed by the virtual wasps Some configurations of cells trigger the construction of a new cell Construction rules are probabilistic • Nest Architectures obtained by simulation

19. Trail recruitment in ants A self-organized behavior based on stigmergic interactions

20. As the ants return from the food source to the nest, they lay down a trail of pheromones that can be followed other ants • Recruited ants lay down their own pheromone on the trail as well, reinforcing the pathway • Trail formation results from a positive feed-back • Negative feedback results from the evaporation of pheromone or the exhaustion of the food source • Trail’s recruitment system enables efficient decision making

21. The ingredients of self-organization • Positive feedback (amplification): they are simple behavioral 'rules of thumb‘ that promote the creation of structures • Negative feedback: counterbalances positive feedback and helps to stabilize the collective pattern: it may take the form of saturation, exhaustion or competition • Amplification of fluctuations: fluctuations act as seeds from which structures nucleate and grow. Randomness enables the discovery of new solutions • Multiple interactions: enable the stochastic nature of the underlying mechanisms to produce the appearance of large and enduring structures

22. Example :Collective Behaviours in Groups of Insect like Robots Definition: Robotics becomes collective when more than one robot are involved in a task

23. Centralized organization • In a centralized system, one particular agent (the leader) gathers all relevant information and plansthe tasks of the entire team • The centralized approach suffer from the leader’s limits in terms of acquisition, treatment and emission of the information • The acquisition and the emission limits the size of the team and the range of action • The treatment limits the speed at which the system responds to quick changes

24. Decentralized organization • In a decentralized system, each robot will decide its behaviour in a full autonomousway according to the information it gathers • The autonomy of the robots avoids the communication bottleneck introduce by a leader • Therefore, the size of the team and the range of action are theoretically not limited • Organization of each robot’s activities is not straightforward in decentralized systems • Today, no formal method exists to easily design individual robot controllers from a desired collective behaviour

25. Social animals Swarm robotics • Swarm robotics is a term originally coined by Gerardo Beniat the end of the 1980’s • This term refers to methods for the organization of groups of robots based on the principles of swarm intelligence • Swarm intelligence is “any attempt to design algorithms or distributed problem-solving devices inspired by the collective behaviour of social insect colonies and other animal societies”

26. Collective behavior can be understood as the combinationof four functions of organization: • Coordination • Cooperation • Collaboration • Deliberation

27. Coordination The appropriate organization in space and time of the tasks required to achieve specific collective behavior : Wasps use the local configuration of the cells to organize their nest building activity

28. A commercial company made use of this principle to design a multi-robot building system • Two different kinds of robots were involved in this task

29. Collaboration The allocation of the tasks between several groups of specialized agents In animals, the allocation of tasks can depend of physical differences between the workers.

30. This simple principle has been used to design a process of adaptive division of labour for a group of robots engaged in a foraging task

31. Cooperation Cooperation: the achievement by a group of a task that could not be achieved by a single individual

32. Cooperation is a key function in nature since it allows to overstep the limits of the individuals • It is often associated with direct communication between the cooperating individuals • This experiment clearly shows that cooperation can occur even with indirect interactions, here mediated by the force applied on the wood stick

33. Deliberation The collective choice of an opportunity among several spread in the Environment Ants are able to achieve this by their pheromone trail laying : This principle is used in path optimization algorithmsfor telecommunication networks

34. Porting such principle to collective robotics is problematic because dealing with chemical signals is difficult

35. Search and rescue

36. Conclusion and Perspectives • Complex colony-level structures and swarm intelligence of social insects emerge from decentralized interactions among individuals • Swarm intelligent systems are flexibleand robust • Social insects are a rich source of inspiration to design adaptive decentralized artificial systems

37. References • http://en.wikipedia.org/wiki/Gerardo_Beni • http://link.springer.com/article/10.1007%2Fs11721-007-0004-y#page-1 • http://www.theswarmlab.com/ • http://www.swarm-bots.org/