Um arcabouço estocástico para coleta de dados em RSSF utilizando um grupo de robôs móveis cooperativos

Abstract

The aim of this work is to propose an approach to multi-robot cooperation in data collection tasks, considering actuation and observation errors, limited communication and uncertainty about sensor node position. The methodology consists in two types of robots: 1) Task Robots that performthe designed tasks and 2) Scout Robots that explore the environment. The cooperation among the robots is modeled with Decentralized Partially Observable Markov Decision Process. The use of these robots is investigated analyzing several metrics, such as Expected Reward. This approach has been applied to several scenarios with sensor nodes deployed in unknown positions. Experiments were conducted to validate the proposed methodology.