Survey of Flocking Algorithms in Multi-agent Systems Aabha Barve 1 , Manisha J. Nene 2 1 Department of Applied Mathematics Defence Institute of Advanced Technology, Pune, Maharashtra, India 2 Department of Computer Science and Engineering Defence Institute of Advanced Technology, Pune, Maharashtra, India Abstract Flocking behaviour in Multi-agent Systems (MAS) has attracted tremendous attention amongst researchers in the recent past due to its potential applications in various fields where distributed work environment is desired. The flocking algorithms have the potential to introduce self-organizing, self-healing and self- configuring capabilities in the functioning of a distributed system. The flocking algorithms exploit various artificial intelligence techniques, mathematical potential functions and geometric approaches to realize the global objectives by controlling local parameters. The main parameters of characterization of any flocking algorithm consist of mathematical models of agents, their hierarchical or flat control structures and the control approach by which these agents are controlled to exhibit flocking behaviour along with any type of formational constraints. A rigorous survey study of flocking algorithms for agents in MAS in the perspective of various instances of agents shows that there lies a huge scope for the researchers to apply, experiment and analyse various techniques locally to achieve global objectives. This paper surveys the flocking algorithms in perspective of these parameters. Keywords: Flocking Algorithms, Multi-agent systems, Formation control, Leader-follower flocking, Wireless Sensor Networks, Multi-robot systems 1. Introduction An agent is any living or non-living, virtual or physical computational quantity which demonstrates autonomous behaviour and which is reactive, proactive and has social ability to communicate. Hence the examples of agents include living organisms, robots, sensors, autonomous vehicles and even software programs which have above mentioned properties. Such multiple homogeneous/ heterogeneous agents together form a multi-agent system which can build distributed complex system to solve large scale problems that are distributed in nature. Formally a Multi-agent System can be defined as a collection of a number of agents that 1) interact through communication, 2) act/react in an environment 3) have different “spheres of influence” which may coincide, and 4) are linked by organizational relationships. [1] Multi-agent system approach of problem solving offers several advantages to a system that is distributed in nature such as flexibility, robustness, reliability, efficiency and speed, maintainability, reusability and reduced cost. Flocking is one of the behavioural properties of multi- agent systems. Flocking is a form of collective behaviour of large number of interacting agents with a common group objective [2]. Reynolds introduced three heuristic rules that led to the creation of the first computer animation of flocking, which are termed as flocking rules of Reynolds [3]. The important aspects of these flocking rules represent the cohesion, separation and alignment features of a flock, described as Flock Centering, Obstacle Avoidance and Velocity Matching respectively. Amongst the multiple agents in a flock, a Flock Centering rule attempts to let an agent stay close to the nearby flock mates. An Obstacle Avoidance rule will guide an agent to avoid collisions with nearby flock mates. The Velocity Matching rule will let the agents in the flock to match velocity with nearby flock mates. Figure 1 presents the flocking rules which collectively lead to flocking behaviour. Figure 1: Flocking Rules IJCSI International Journal of Computer Science Issues, Vol. 10, Issue 6, No 2, November 2013 ISSN (Print): 1694-0814 | ISSN (Online): 1694-0784 www.IJCSI.org 110 Copyright (c) 2013 International Journal of Computer Science Issues. All Rights Reserved.