The widespread adoption of robotic manipulators across industrial, medical, and service applications has created growing demands for high-precision trajectory tracking control. However, flexible-joint manipulator (FJM) systems in practical operation are vulnerable to actuator failures, resulting in degraded control performance. To address these issues, this paper presents an adaptive neural network (ANN) control algorithm that compensates for system uncertainties and actuator faults through neural networks’ adaptive capability, with closed-loop system stability rigorously proven via Lyapunov theory. Experimental results on the Baxter manipulator demonstrate the proposed method’s superior disturbance rejection and trajectory tracking accuracy compared to conventional approaches, providing an effective solution for high-performance control.