An Exploration Algorithm for a Swarm of Homogeneous Robots

Abstract

We describe a distributed algorithm for coordinating a swarm of homogeneous robots to explore an unknown area using only local interaction. The basic movement rule is based on the gas dispersion model. To avoid duplication exploring, we introduce a new strategy called SMARK to leave signpost in explored area. Consequently this system have three outstanding features: (1) The swarm of robots coordinate to adapt themselves to different environments; (2) The process is independent to the number of robots and can automatically adapt to the influx and death of robots; (3) Only local communication is used, ensuring a low communication traffic. Our experiments were conducted in simulation.We compare different robot group size with up to 70 robots, and evaluate the performance to two forms of environments.The result shows in both environments, the increase of group size will evidently improve system efficiency, and the SMARK plays a more important role in the environment with more obstacles.