A parallel shape formation method for swarm robotics

Abstract

In order to cope with diversified tasks and unstructured environments, a parallel shape formation method for swarm robotics is introduced to adjust the system’s configuration autonomously and flexibly with the user-specified 2-D shape. Given the desired shape to be formed in the form of analytic functions, the forming task of the swarm system is divided into two sub-tasks: the edge robot forming task and the internal robot forming task. Then, the seed robot is selected through the generation and transmission of the gradient to establish the relative coordinate system, and the initial coordinates of robots are obtained through trilateral positioning. Based on that, using the artificial potential field method, under the action of gravitational force generated by the objective analytic functions and repulsion forces generated by other individuals in the neighborhood, edge individuals move to the target boundary; at the same time, internal individuals spread evenly in the target area under the action of the repulsion forces generated by each other. During the forming process, the two sub-tasks are executed in parallel, and the individuals continue to update their real-time positions by dead-reckoning until the desired shape is formed. We evaluate the feasibility and scalability of this novel method in simulation-based experiments, and implement the parallel shape formation algorithm on the Cilibot robot, a hardware system developed in our lab.