LEADs: A Swarm Behavioral Decision-making System for Different Tasks

Abstract

Swarm robots perform better at complex tasks than single robots because they have exceptional performance that solo robots do not. The behavioral decision-making of swarm robots, in particular, is a very important study area that still has a lot of unresolved problems, such as how to integrate the existing swarm behaviors to adapt to different kinds of jobs. In order to achieve a unified control framework for swarm robots, which can be commanded to perform tasks by simply changing the fitness function, this study proposes a system for generating swarm behavior control strategies. The system creates a behavior library by combining the current swarm behaviors and using the K-means clustering algorithm to reduce the dimension of the swarm state space. Then, find a better solution by the greedy algorithm that can finish the job. The experimental findings demonstrate that the system not only scales better in terms of the number of swarm robots but also more effectively combines basic behavior, giving the swarm a high degree of adaptability and scalability across a variety of tasks.