The main objective of this research is to develop practical recurrent neural network-based controllers that can be implemented on a robotic manipulator in the presence of dynamic uncertainties in the system or external disturbances caused by end-effector payload variation. A hierarchical intelligent control scheme was proposed, and two recurrent neural network-based controllers were designed: i) a Second-Order Recurrent Neural Network (SORNN)-based controller; and ii) a Bistable Gradient Network (BGN)- based controller. The SORNN-based controller was developed to guarantee the trajectory tracking control performance as well as the stability of the overall controller and robot system. Extensive experiments of the SORNN-based controller were conducted to validate the expected control performance improvement. BGN was designed for system identification and predictive control of 2 and 3 DOF robots. Simulation results demonstrated BGN’s capability of compensating for nonlinear disturbances. Experiments could not be performed, but the analysis of BGN’s applicability was given.