Abstraction based approach for segregation in heterogeneous robotic swarms

Abstract

The focus of this study is to design individual control laws that segregate multiple groups of mobile heterogeneous robots. Our approach is based on the use of abstractions to represent each group of robots and an artificial potential function to segregate the groups. Different from other works in the literature, we prove that with our controller the system will always converge to a state where robots of the same group will be together while separated from robots of different groups. We also propose a collision avoidance scheme which does not interfere in the segregation controller. Furthermore, our controller has a local property, meaning that the controller might not require global information of the whole swarm to converge to the segregated state. The approach is validated with simulations varying the number of robots and groups and experiments with real robots.