To address the severe inter-axis cross-coupling and positioning accuracy limitations in three-dimensional piezoelectric nano-positioning platforms during multi-axis motion that hinder nanometer-level measurement requirements, this study proposes a hybrid control algorithm integrating feed-forward decoupling control theory with sliding mode control. The method involves: first, establishing a coupled error model and employing least squares estimation for parameter identification to design feed-forward decoupling controllers that separate the coupled system into independent single-axis control units; then, developing a non-singular terminal sliding mode self-stabilizing controller to mitigate piezoelectric hysteresis nonlinearity, residual coupling effects, and external disturbances, while utilizing extended state observers for real-time disturbance estimation and compensation, and using exponential convergence laws to suppress sliding mode chattering. The effectiveness of the control algorithm was verified through MATLAB / Simulink simulation, and an experimental platform based on a laser interferometer was constructed for coupled testing of a three-dimensional piezoelectric nano-positioning system. Experimental results demonstrate that the proposed sliding-mode decoupling control algorithm reduces the error caused by inter-axis coupling to below 0.03 μm, with the grid scanning trajectory closely matching the planned path, thereby meeting the nanometer-level positioning accuracy requirements of the multi-axis piezoelectric platform.