A new fuzzy sliding mode controller is designed for the chattering problem inherent in the sliding mode control algorithm. Firstly, the Lagrangian method was used to establish the dynamic model of the double joint manipulator. Then, based on the Lyapunov stability principle, according to the sliding mode arrival condition, the fuzzy rule was used to blur the switching gain and to replace the original switching gain. The new fuzzy sliding mode controller was finally designed and simulated with MATLAB/Simulink and compared to a normal sliding mode controller. Compared with the original controller, the fuzzy sliding mode controller still had a good effect on the chattering in the case of external nonlinear interference. It has better control performance, better tracking effect and faster spee, responding speed. The research done has a certain reference to the motion control of the actual robot arm.