In this paper, we aim to improve the tracking performance of the manipulator joint system under the presence of the uncertainties, such as modelling error, friction, and external disturbance. Firstly, the nonsingular fast terminal sliding mode control is developed to guarantees a finite-time convergence and to solve the singular issue of the terminal sliding mode control. Secondly, in view of the established system model, an adaptive sliding mode controller (SMC) based on radial basis function neural network (RBFNN) and sliding mode variable structure control theory is designed for the tracking of the bi-joint manipulator and six-degree of freedom parallel robot. Finally, the results show that our method improves the robustness of the adaptive RBFNN controller further, weakens the chattering phenomenon, reduces error, and has an excellent control performance.