Distributed Formation Control for Second-Order Nonlinear Multiagent Systems Using Predictor-Based Accelerated Fuzzy Learning

This paper investigates the distributed formation control problem of second-order multiagent systems (MASs) subject to unknown nonlinear dynamics. A predictor-based fuzzy learning framework is developed, in which fuzzy logic systems (FLSs) approximate the unknown nonlinear dynamics and a two-layer learning mechanism accelerates weight adaptation by utilizing the velocity prediction error. A distributed observer is designed for jointly connected switching directed topologies to handle limited leader accessibility, allowing agents to estimate the global reference trajectory using solely local information. Furthermore, an actor–critic reinforcement learning architecture is integrated to realize distributed optimal formation control, in which the value function is reformulated with a compensatory term that exploits available system structural information, thereby enhancing policy optimization efficiency and accelerating convergence. Simulation results demonstrate that the proposed approach exhibits improved robustness and accuracy in both dynamics approximation and position tracking when compared with conventional learning-based methods.