Abstract: To address the tracking control challenges of robotic arms faced with full state constraints, bounded unknown disturbances, and dynamic uncertainties, we propose an adaptive neural network control strategy that employs tangent-type barrier Lyapunov functions to manage full state constraints and bounded disturbances. Time-varying constraints are applied to position errors, while static constraints handle velocity errors. A time-varying class PD term is introduced in the virtual control design to speed up system response. To address and suppress the bounded disturbances caused when the end of the robotic arm carries an object without fixing it well, an adaptive neural network is used. This approach effectively deals with system uncertainties, ensuring that the robotic arm satisfies predefined state constraints even under external disturbances and unknown dynamics. The Moore-Penrose inverse and Lyapunov stability theory are introduced to prove that the closed-loop system remains consistently bounded. Comparative simulation results demonstrate the method's advantages in achieving fast response speeds, small tracking errors, and strong robustness to full-state constraints. Experimental results on a Franka Emika Panda robot validate the effectiveness of the proposed method.