Assessing the value of coordination in mobile robot exploration using a discrete-time Markov process

Abstract

With the introduction of multi-robot systems for exploration the question then arises, whether coordination among robots in such systems is required. We propose a model based on Markov processes to evaluate the need for coordination in multi-robot systems during the exploration of unknown environments and determine possible gains achievable through coordination. The model is illustrated by exploration of an indoor office environment. We qualitatively identify characteristics of environments which make coordination necessary and allow to quantitatively include them in the model. The expected gain through coordination highly depends on the environment. We further investigate the impact of team sizes. In favorable environments explicit coordination is not needed at the cost of increased team sizes. This helps to raise understanding of factors having an impact on coordination functions and making it possible to approximate a possible gain through coordination.