In this paper, we consider a persistent coverage control problem for a group of two-wheeled mobile robots in an environment with unknown obstacles. The persistent coverage control is one of the multi-agent control method which provides efficient and persistent environmental monitoring. In typical persistent coverage control, the control inputs are calculated based on a given time-varying density function designed for monitoring and might not work well with unknown obstacles. For the coverage control in the unknown environment, we propose a persistent coverage control method utilizing the prediction of unknown obstacles by Gaussian process regression. The features of the proposed method is discussed based on simulation and experimental results.
Ishihara et al. (Tue,) studied this question.