This chapter presents an adaptive impedance control approach to facilitate active rehabilitation using a wearable robot with (7-DOFs) with an unknown dynamic model. A key challenge in this approach is enabling the exoskeleton to provide precise assistance by accurately interpreting the wearer’s movement intentions. Due to human movement’s nonlinear and time-varying nature, determining the user’s desired motion intention (DMI) is complex. To address this problem, we use Hill’s model to estimate the human forces using the gravity of the human arm, force sensors, and electromyogram (sEMG) signals. A Radial Basis Function Neural Network (RBFNN) is then employed with a sliding mode estimator to estimate the DMI in real time. This estimated DMI is integrated with the proposed adaptive impedance control to ensure the robot can track a specified impedance target. As a result, the proposed control strategy enables the exoskeleton to predict the desired trajectory, providing comfortable and adaptive assistance. Experimental trials with healthy subjects under various scenarios confirm the effectiveness of this approach in achieving smooth collaboration between the exoskeleton and its wearer.