Inspired by social biological organisms, a Swarm Robotics System (SRS) has been proposed as a type of multi-robot system that consists of many homogeneous autonomous robots. The source seeking behavior of social insects would have numerous applications to a robotic swarm in the real world, such like search and rescue. The main objective of this work is to solve the problem of source localization by a SRS with communication and payload constraints. We used the Particle Swarm Optimization (PSO) method to describe the final probable source position as a local property of an agent, combined with Mean Embeddings (ME) based local observation of the agent as inputs of an Artificial Neural Network (ANN). We demonstrate that this method enables to solve the source localization and the swarm behavior can be controlled by changing the PSO parameters.
展开▼