The hydraulic-driven lower limb exoskeleton robot (HDLLER) can provide excellent assistance during human walking. However, complex torque coupling disturbances exist between each joint, negatively impacting the precise torque tracking of each joint channel of the robot. To address the coupling force disturbances between HDLLER joints and the human-robot interactions, this paper proposes a virtual torque control (VTC) strategy based on modal decoupling. Specifically, a human-robot coupled dynamic model of the HDLLER considering human motion disturbances is first established. Then, based on vibration theory, a modal space decoupling approach is proposed to transform the system's mass and stiffness matrices into diagonal matrices, creating two independent control channels. Furthermore, a VTC strategy is introduced to compensate for disturbances caused by human motion and the residual terms after modal decoupling, thereby enhancing the HDLLER's performance. Finally, to handle parameter variations during modal decoupling and inaccuracies in model identification, the H