Dispersion and Line Formation in Artificial Swarm Intelligence

One of the major motifs in collective or swarm intelligence is that, even though individuals follow simple rules, the resulting global behavior can be complex and intelligent. In artificial swarm systems, such as swarm robots, the goal is to use systems that are as simple and cheap as possible, deploy many of them, and coordinate them to conduct complex tasks that each individual cannot accomplish [Cao 1995; Simmons 2000; Sahin 2005]. The system may involve a group of artificially intelligent agents with limited sensing, communication, and computing capabilities and can achieve a higher level task by the agents locally or globally collaborating with each other. Local collaboration can be established among the neighbors within a certain communication range or physical/social distance. Global collaboration can be realized by centralized control possibly with a designated leader. Global shape formation is one of the challenging problems in artificial swarm intelligence. In nature, it is performed for various purposes, such as engergy saving and path optimization. For instance, a flock of large birds fly together while forming a V shape in order to reduce the air resistance and fatigue for physically weak birds [Weimerskirch 2001]. Ants form a line to optimize the path to a food source by laying a trail of pheromone and regulate foraging activities without any central control or spatial information [Prabhakar 2012; Gordon 2013]. This trail enables the ant colony find the shortest path to the food source [Bonabeau 2000]. Shape formation in artificial intelligence systems is usually required for specific task-oriented performance, including 1) forming sensing grids [Spears 2004], 2) exploring and mapping in space, underwater, or hazardous environments [Nawaz 2010; Wang 2011], and 3) forming a barricade for surveillance or protecting an area/person [Cheng 2005]. This paper presents a dynamic model of an artificial swarm system based on a virtual spring damper model and algorithms for dispersion without a leader and line formation with an interim leader using only the distance estimation among the neighbors.