A Data-Driven Algorithm Based on Neural Network Prediction for Agent Trajectory Tracking

A distributed model-free adaptive control scheme for agent systems is presented in this paper, which solves the trajectory tracking problem in fixed communication topology. The unknown nonlinear function is used to model the dynamic equation of each agent. Furthermore, the desired trajectory required is acceptable to a portion of the following agent. The main feature of this method is that it can solve the trajectory tracking control problem only by relying on the I/O data of each agent. A neural net-work is used to estimate the unknown quantity in the adaptive control law. Theoretical analysis shows that the tracking error is convergent and the output and control input sequences are bounded. Finally, a simulation is given to verify the effectiveness of the proposed design method.