Abstarct: This thesis proposes novel approaches to optimize the impedance control of lower limb rehabilitation robots. Since the rehabilitation robots are completely in contact to the patient's body, they are subjected to uncertainty. Therefore, on the one hand, there is no exact model for the whole system and on the other hand, the dynamic behavior of this robot is of great importance and sensitivity. To compensate for the uncertainty, the first proposed impedance control method uses the output of an expected internal model and predicts the uncertainty in a finite horizon. In the second proposed method, to adjust the parameters of the proposed impedance controller, a bi-level optimization problem is designed, by solving which, the parameter adaptation law and the optimal adaptation gain are obtained, simultaneously. It should be noted that the existed adaptive methods do not pay attention to system stability and optimal parameter adaptation, at the same time, however, in the proposed structure, this is achieved through the bi-level optimization. The third proposed method is to use the offline method to optimize the controller’s parameters, baxsed on the voltage control strategy, which is not already possible in the conventional methods. Moreover, a specific desired trajectory is designed for rehabilitation robots, which fully covers the practical considerations related to rehabilitation, and provides a smooth path with sufficient rest on both ends and an analytical equation with the ability to adjust the acceleration and speed thresholds. The comparison studies confirm the superiority of the proposed methods over similar control methods. Finally, in this research, the proposed control methods have been successfully implemented for the first time on a knee joint rehabilitation robot designed and built by the student.