Sensor noise significantly degrades the trajectory tracking accuracy of robotic manipulators. In this paper, we propose a closed-loop control strategy based on multisensor fusion and an extended Kalman filter (EKF) to enhance performance under noisy sensing conditions. A six-degree-of-freedom Puma 560 manipulator is modeled, including its kinematics, dynamics, and stochastic measurement models for encoders and an inertial measurement unit. The EKF fuses multisensor data to provide real-time estimates of joint positions and velocities, which are integrated into a computed torque controller. Comparative simulations in Matrix Laboratory/Simulink demonstrate that, compared with conventional feedback using direct noisy measurements, the proposed approach reduces end-effector position error, suppresses velocity estimation fluctuations, and achieves smoother trajectory tracking. We validate the effectiveness of EKF-based sensor fusion for improving the robustness of robotic manipulators in noise-affected environments. All results are obtained from simulations and have not been validated on physical hardware; sensor non-Gaussian noise, sudden faults, and joint flexibility are not considered. Future work will focus on hardware-in-the-loop validation, adaptive robust filtering, and deep-learning-assisted noise modeling to address these limitations.