Observer-based adaptive emotional command-filtered backstepping for cooperative control of input-saturated uncertain strict-feedback multi-agent systems

This paper introduces a distributed observer-based emotional command-filtered backstepping (DOECFB) approach for leader-following cooperative output-feedback control of heterogenous strict-feedback multi-agent systems (MAS) under mismatched uncertainties and input saturation. A novel state observer is designed based on radial-basis emotional neural networks (RBENNs) that approximate uncertainties of model dynamics. To model inter-agent dynamics with less complexity, emotion-inspired approximated dynamics are shared among neighbouring followers, like emotional contagion in a group of people. An auxiliary system is also used to attenuate input saturation's negative effect on the cooperative tracking performance. Also, command filters and compensating signals are applied to avoid the 'explosion of complexity' in the backstepping design. Only local information from other agents is required for the proposed approach to guarantee convergence of the cooperative tracking error to a small region around zero and cooperatively semi-globally uniformly ultimately boundedness of closed-loop signals. Simulation examples on a second-order uncertain MAS and multiple forced-damped pendulums are conducted, and quantitative comparisons verify the effectiveness of DOECFB and the proposed observer.