Robust Neural networks Compensating Motion Control of Reconfigurable Manipulator in Geometric Form

There are many uncertainties in real dynamics system of reconfigurable manipulator that makes PID etc. traditional methods control imprecisely. Thus, robust neural networks compensating control scheme is developed to compensate structured and unstructured uncertainties to enhance computed torque control based method. Kinematics is described by POE formula and the dynamics are derived by Newton-Euler algorithm of geometric form. They are better on computation aspects and more universal on n-links model than other modeling methods for reconfigurable manipulator system proved by many researchers. Under some mild assumptions, a stability analysis guarantees that tracking errors are uniformly ultimately bounded. Finally, simulations of the proposed controller on a RRP reconfigurable manipulator demonstrate the effectiveness on satisfactory tracking performance