Distributed Critic-Based Neuro-Fuzzy Learning in Swarm Autonomous Vehicles

Drawing from the latest breakthroughs in swarm robotics and smart control systems, our study explores frequency synchronization and phase alignment in oscillatory networks. The core of our work is a novel distributed consensus protocol paired with a reinforcement learning algorithm tailored for a leader-follower multi-agent oscillatory network. We present a critic-based neuro-fuzzy learning strategy designed to unify phase and frequency tracking while reducing errors. Each agent is tasked with maintaining its assigned phase and frequency. Our method utilizes a fuzzy critic to assess situations and adjust the controller’s parameters accordingly, aiming to diminish the stress signal. Our versatile design adapts to various networks, countering dynamic agent behaviors and network parameter uncertainties, ensuring reliable controller operations. Our scalable solution supports numerous autonomous agents. We demonstrate our learning strategy’s efficacy through simulations on a network of ten oscillating vehicles. Furthermore, we conducted a comparative analysis of the results obtained from the implementation of the proposed approach against those derived from the use of a conventional PI controller.