Funnel Neural Network Model Predictive Control for a 4 DoF Robot Manipulator

This paper introduces the design of a Prescribed Performance Neural Network Model Predictive Control (PPNNMPC) tailored for a 4 Degrees of Freedom (DoF) robot manipulator in trajectory tracking applications. The design of the proposed controller is formulated using Model Predictive Control (MPC) methodology, which has the advantage of predicting the system's future behavior to achieve optimal robot control. Then, the prescribed performance function is incorporated into the control law by integrating the transformed error in the optimization process. The Prescribed Performance Function (PPF) maintains the tracking error within predefined limits, enhancing the system's transient response. Furthermore, integrating Neural Networks (NN) and prescribed performance functions into the control law design mitigates the common computational time drawback associated with MPC. Simulation results emphasize the superior efficiency of the suggested controller in contrast to the traditional model predictive control, demonstrating reduced overshoot, small settling time, improved computational efficiency, and accurate tracking of desired set-points.