A strategy based on Wave Swarm for the formation task inspired by the Traveling Salesman Problem

Multi-robots can perform complex tasks such as exploration, foraging, and formation. Efficient communication between robots can contribute to the accomplishment of collective tasks through efficient message exchange. This paper proposes a strategy based on the message’s propagation technique, Wave Swarm, for the formation task inspired by the Traveling Salesman Problem. The Wave Swarm communication approach uses the concept of wave propagation for message exchange between neighbors, establishing a Father and Son relationship between robots. However, different pairs of Father and Son robots can impact the simulation time, the average distance traveled by each robot, and the number of messages exchanged during the formation task. Thus, given a random distribution of robots into a swarm, we model the problem of assigning one position and route for each robot to achieve its place in the formation as a Traveling Salesman Problem. The routes resulting from the TSP solution establish a new parental relationship between the robots in the swarm. We performed preliminary experiments to define the technique used in the TSP resolution. We tested reinforcement learning and the genetic algorithm using the TSPLIB95 library. Thus, we develop a strategy for formation tasks based on Wave Swarm and TSP solved with reinforcement learning. We adopted the leader–follower approach in an unknown environment to validate the proposal. The results show the behavior of different sizes of robot groups for various desired shapes. Experiments with the robot simulator CoppeliaSim (V-REP) validate the proposed strategy and highlight its efficiency and robustness while running the formation task.