چكيده به لاتين
Abstract
Control of rehabilitation robots have been very active field during the past two decades and even before it. Rehabilitation robots have several advantages over conventional rehabilitation that has attracted the attention of researchers. This includes, reducing the number of physical therapy from a few people to one person, rehabilitation action is more efficient and reduces the period of rehabilitation. The control of lower-limb rehabilitation robots, which is the subject of this thesis is more difficult than robots that are not in direct contact with humans since these robots are in direct interaction with the patient and hence optimum performance of these robots in order to maintain patient safety and also to advance the goals of rehabilitation is very important. In addition to the motion control, control stability is also discussed here. The motion control of lower-limb rehabilitation robot with five degrees of freedom using non-linear model predictive control for walking on the treadmill and flat surfaces is the aim of this thesis. In this method, the pattern of robot movements is implemented implicitly and in real time in the model predictive control method and by considering the dynamic forces applied from the human and the human-equivalent torque in the dynamic equations of the robot, the human dynamic is implemented. By designing the movement pattern using the model predictive control, the optimal paths for the robot joints moments by considering the practical limits are generated. The main advantage of using the predictive controller is to obtain optimum torque and taking into account the constraints of the rehabilitation robot movements. Moreover, the model predictive control method can reject external disturbances and provide optimal performance in presence of the measurement noises. In this thesis, three steps were considered for rehabilitation of lower limbs. The first step is at the beginning of rehabilitation procedure. The second and third steps are considered for the mid-term and final stage of the rehabilitation process, respectively. At each step, according to the conditions of the corresponding step, the patient's will can be observed and considered in his/her rehabilitation process. According to the rehabilitation phase, the gait length and height, and in general, the movement pattern that are incurred by the patient are considered for the rehabilitation. The external disturbances are rejected very well. Moreover, using the Poincare mapping the robot's stability is analyzed. The simulating results show better performance of the proposed method as compared with the previously reported methods.
Key words: Lower limb rehabilitation robot, Nonlinear model predictive control, Human-robot interaction, Poincaré mapping
