The dynamic model of the two-link robot manipulator based on LuGre friction is established in order to solve the problem that the control precision of manipulators is difficult to be improved of because of unknown nonlinear friction. An adaptive fuzzy-neural-network controller(AFNNC) is designed subject to the high-coupled and nonlinear of the system and numerous parameters of the LuGre friction. Fuzzy-neural-network(FNN) compensator is used to approximate the nonlinear part including the LuGre friction. The center and the width of the basic functions of the FNN can be fixed with on-line adaptation, and the approximate error is eliminated by a sliding-mode part. The stability of the close-loop system is proved by using Lyapunov function. Simulation results show the effectiveness of the control algorithm.