Distributed predefined-time neural adaptive control design for consensus of networked Euler-Lagrange systems

This article presents a distributed predefined-time neural adaptive control scheme for the consensus of networked Euler-Lagrange (EL) systems under model uncertainties and external perturbations through directed communication topology. First, a distributed predefined-time observer is constructed to estimate the leader's state information for each follower agent. Then, based on the recovered information, the predefined-time local controller is designed for each follower agent by utilising the predefined-time backstepping control approach. Moreover, the neural network (NN) is incorporated to identify the lumped unknown item. Particularly, an indirect learning mechanism is adopted to determine the upper bound of the optimal NN weight. In this way, the computational cost of the presented controller is greatly degraded. Stability evaluation shows that the presented controller can guarantee the position and velocity synchronisation errors regulate to the minor regions around zero in predefined time. Lastly, simulated studies are conducted on the consensus of networked robotic manipulators to validate the obtained results.