Self-organized Rectangular Aggregation of Swarm Robotics with Minimum Sensing Range

In self-organized aggregation, that no specific aggregated sites are known makes it intractable for powerless robots. In this paper, we formulate the aggregation problem as a shape construction process in which agent's behavior is determined by the positional relationship with its neighbors. This paper presents a novel self-organizing strategy to gather a swarm of homogeneous robots into rectangular clusters without any prior target knowledge. Moreover, the robots are unable to communicate with each other and have a minimal range of perception, which is the smallest value to avoid collision between individuals. By the definition of semi-stable/stable states, we design a distributed probabilistic aggregation algorithm that takes advantage of the temporal and spatial perception of robots. A new metric is introduced to measure the similarity of aggregated clusters and rectangles. This algorithm is evaluated via a series of simulation experiments running on a group of up to 1000 agents, showing that it allows robot swarms to obtain the desired aggregation.