A nonlinear disturbance observer (NDO) is proposed for robotic manipulators to estimate and compensate for disturbances such as friction, unmodeled dynamics, and reaction forces. The NDO is designed to overcome the limitations of existing linear disturbance observers (DOs), which are based on linear system techniques. The NDO is derived using nonlinear methods and is guaranteed to achieve global exponential stability by selecting appropriate design parameters based on the maximum velocity and physical parameters of the manipulator. The observer is applied for various purposes, including friction compensation, independent joint control, sensorless torque control, and fault diagnosis. The performance of the NDO is demonstrated through simulation and experimental results for a two-link robotic manipulator. The NDO is shown to effectively estimate friction, even in the presence of rapid variations and discontinuities. The observer is implemented on a two-link robotic manipulator, and experimental results show improved tracking performance and reduced steady-state error when the estimated friction is used for feedforward compensation. The NDO is also applicable to other robotic applications such as independent joint control and sensorless torque control. The paper presents a systematic method for designing the NDO and establishes conditions for its convergence. The design parameters are chosen to ensure global stability, and the convergence rate can be specified by the design parameters. The NDO is shown to be effective for both constant and rapidly varying disturbances, demonstrating its robustness and versatility in robotic control applications.A nonlinear disturbance observer (NDO) is proposed for robotic manipulators to estimate and compensate for disturbances such as friction, unmodeled dynamics, and reaction forces. The NDO is designed to overcome the limitations of existing linear disturbance observers (DOs), which are based on linear system techniques. The NDO is derived using nonlinear methods and is guaranteed to achieve global exponential stability by selecting appropriate design parameters based on the maximum velocity and physical parameters of the manipulator. The observer is applied for various purposes, including friction compensation, independent joint control, sensorless torque control, and fault diagnosis. The performance of the NDO is demonstrated through simulation and experimental results for a two-link robotic manipulator. The NDO is shown to effectively estimate friction, even in the presence of rapid variations and discontinuities. The observer is implemented on a two-link robotic manipulator, and experimental results show improved tracking performance and reduced steady-state error when the estimated friction is used for feedforward compensation. The NDO is also applicable to other robotic applications such as independent joint control and sensorless torque control. The paper presents a systematic method for designing the NDO and establishes conditions for its convergence. The design parameters are chosen to ensure global stability, and the convergence rate can be specified by the design parameters. The NDO is shown to be effective for both constant and rapidly varying disturbances, demonstrating its robustness and versatility in robotic control applications.