《Design and Assessment of Sweep Coverage Algorithms for Multi-agent Systems with Online Learning Strategies》

Abstract: Cooperative sweep coverage of multi-agent systems has found broad applications in various fields. This paper proposes a scheme to address the sweep coverage problem of multi-agent systems (MASs) within uncertain environments. In the proposed formulation, the coverage region is divided into multiple stripes, of which each has the workload completed by MASs in sequence. When the workload on the current stripe is completed, all the agents switch to the next together. The temporal dependence between the switching time computation and the sweep coverage operation is taken into account, and an online learning strategy is designed to handle environmental uncertainties and balance the workload among agents on the same stripe. Thereby, the distributed sweep coverage algorithm is developed to guarantee the complete sweep coverage, which consists of three operations, i.e., communication, workload partition and sweeping. Theoretical analysis is afterwards conducted to estimate the upper bound for the error between the actual and optimal coverage time. Finally, numerical simulations are carried out to substantiate the effectiveness and superiority of the proposed scheme. 

Index Terms: Distributed control, distributed estimation, learning control systems.