Swarm robotics is the study of how large numbers of relatively simple physically embodied agents can be designed such

that a desired collective behavior emerges from the local interactions among agents and between the agents and the environment (Sahin, 2005). Dorigo and Sahin (2004), complain of the plethora of terms with often vague and overlapping meanings that are applied to multi-robot research, and use their editorial introduction to a special issue on Swarm Robotics to propose a set of defining criteria. They suggest that swarm robotics studies should (i) aim for scalability and to be relevant for the coordination of large numbers of robots (ii) should consist of a small number of homogenous groups, each containing a large number of members; (iii) should be based on robots that are ill-equipped to carry out tasks on their own and that perform better when they cooperate and (iv) that they should have only local and limited sensing and communication ability. Sahin (2005) extends these criteria to include the proviso that the robots themselves should be autonomous intro

Swarm robots are simple and cheap robots with limited sensing and computational capabilities. They are feasible to deploy teams of swarm robots and take advantage of the resulting fault tolerance and
parallelism

Swarm robots cannot use conventional planning methods due to their limited sensing and computational
capabilities

The theme of simplicity and elegance resonates throughout swarm robotics research in both the designs of
the robots as well as the algorithms that are devised for these systems. adv

Robot swarms are able to cover more area than an individual robot
Distributed computing is becoming more and more prevalent in engineering today. Swarm robotics is simply an extension of that, not only dividing the computing power, but also the physical capabilities. Fault tolerance De-centralized approach Flexibility Robustness

Another feature of robot swarms is their effectiveness scales well with the number of members. The
algorithms for swarms scale well and do not depend on the number of robots.On the other hand, it is not always clear how to improve the effectiveness of an individual robot system. Often times improvements in hardware require additional software upgrades, which is not the case with swarms. Simple and elegant The robot controller that dictates the behavior of the individual robots is very simple. The behaviors of the individual robots can usually be represented as a state machine with few states and edges. Scalable Swarm robotics algorithms are designed so that they work for any number of robots. Also, they are expected to scale well as new robots are added. Decentralized The robots in a swarm are autonomous and do not follow any exterior commands. Although a member of a swarm can be directly and predictably influenced by the behavior of another, the choice is under its own accord. Being decentralized is often coupled with being scalable. Usage of local interactions Local interactions are used over broadcasting messages in the majority of these algorithms. Even broadcasts are implemented as message-hopping protocols. This ideal is a major factor in the scalability of the system In biological systems such as swarms of bees, or colonies of ants, sophisticated behaviour emerges as a consequence of the interactions of large numbers of individually simple units, that are only capable of local communication, and that operate independently and autonomously.

Aggregation problem requires aggregation of a randomly placed robots in an environment. The problem is easy when a centralized control approach is used but the problem is not trivial when the distributed control is used. Swarm robots are simple and cheap robots with limited sensing and computational capabilities. This makes it feasible to deploy teams of swarm robots and take advantage of the resulting fault tolerance and parallelism. Swarm robots cannot use conventional planning methods due to their limited sensing and computational capabilities. These systems are used to develop useful macro-level behaviors while being made of modules that are very simple in design and compact in size.The theme of simplicity and elegance resonates throughout swarm robotics research in both the designs of the robots as well as the algorithms that are devised for these systems. For example, breadth-first search is extremely inefficient when robots use it. A robot would have to backtrack and travel to that node instead of being able to move a pointer from node to node in a graph. For these reasons and others, devising algorithms for robot swarms has become a research field of its own. Swarms of robots must be built or simulated to test these algorithms, demonstrate they work and improve on previous work Communication Interaction via sensing Interaction via environment (Stigmergy ) Interaction via communication Behavior Nonadaptive Subsumption Probabilistic Finite State Automata Distributed Potential Field Methods Neural Networks ( as a generalization mechanism and do not use its learning capabilities) Learning Reinforcement Learning Local Reinforcement Global Reinforcement Evolution Modeling Sensor-based non-physical simulations physicalsimulations Hierarchy is more efficient but Lack of Hierarchy means: no need of sophisticated abilities no need for map of environment no risk of fatal breakdown ( no one unit being essential ) no need for supervisoin no global communication able to work in large environment cheaper and more robust Dynamic sensor networks are critically needed for various tasks such as search and rescue, surveillance, perimeter defense, locating and mapping of chemical and biological hazards, virtual space telescopes, automated assembly of micro-electromechanical systems, and medical surgery (e.g., with nanobots)