Adaptive Reinforcement Learning Formation Control Using ORFBLS for Omnidirectional Mobile Multi-Robots

This paper presents an adaptive reinforcement learning formation control method using output recurrent fuzzy broad learning systems (ORFBLSs) for cooperative consensus formation of uncertain heterogeneous omnidirectional mobile multi-robots. Having modeled any uncertain heterogeneous omnidirectional mobile robot by using a unified uncertain three-input three-output second-order state equation, we combine the actor-critic reinforcement learning, integral terminal sliding-mode control (ITSMC), and ORFBLS online learning to propose an adaptive ORFBLS-based reinforcement learning formation control, abbreviated as ORFBLS-ARLFC, method. Via Lyapunov stability theory, the ORFBLS-ARLFC method is shown to accomplish asymptotical consensus in the presence of uncertainties. One comparative simulation and one experiment are carried out to exemplify the effectiveness and merits of the proposed ORFBLS-ARLFC method on cooperative omnidirectional mobile robots with uncertainties.