Search

This chapter discusses the advantages and feasibility of using compliant actuators in exoskeletons. We designed compliant actuation for use in a gait rehabilitation robot. In such a gait rehabilitation robot large forces are required to support the patient. In case of poststroke patients only the affected leg has to be supported while the movement of the unaffected leg should not be hindered. Not hindering the motions of one of the legs means that mechanical impedance of the robot should be minimal. The combination of large support forces and minimal impedances can be realised by impedance or admittance control. We chose for impedance control. The consequence of this choice is that the mass of the exoskeleton including its actuation should be minimized and sufficient high force bandwidth of the actuation is required. Compliant actuation has advantages compared to non compliant actuation in case both high forces and a high force tracking bandwidth are required. Series elastic actuation and pneumatics are well known examples of compliant actuators. Both types of compliant actuators are described with a general model of compliant actuation. They are compared in terms of this general model and also experimentally. Series elastic actuation appears to perform slightly better than pneumatic actuation and is much simpler to control. In an alternative design the motors were removed from the exoskeleton to further minimize the mass of the exoskeleton. These motors drove an elastic joint using flexible Bowden cables. The force bandwidth and the minimal impedance of this distributed series elastic joint actuation were within the requirements for a gait rehabilitation robot.

The goal of the work presented in this thesis was to realize a robotic device that is able provide suitable gait training to stroke patients. It is believed that motor training in general, but specifically for stroke patients should be intensive and task-specific in order to reach optimal outcome. Meanwhile, the training of severe stroke patients has proven to be physically very demanding to therapists. For this reason it is believed that robotic aids can be useful alternative for physical therapists to provide gait training. In this research we supposed a training setting in which a therapist is in control of the training, while the robot carries out the physical labor in guiding, assisting and correcting the patient.

We have designed and built a lower extremity powered exo-skeleton (LOPES) for the training of post-stroke patients. This paper describes the philosophy behind the design of LOPES (Fig. 1), motivates the choices that have been made and gives some exemplary results of the ranges of mechanical impedances that can be achieved.

Background: In the past decade, different gait rehabilitation robots have been developed. These robots concentrated on relearning gait by repeatedly enforcing a walking pattern for the whole leg or only for the foot. Such training might be less task specific as generally thought, because learning to walk is more complex than learning a position trajectory. We believe active participation of the patient in training and subdivision of the training in several subtasks leads to more comprehensive and functional training. Methods: Using a newly developed actuated exoskeleton (LOPES) in combination with a treadmill, we aimed to selectively and gently support specific subtasks of walking, like knee stabilization, lateral balance, and foot clearance. Foot clearance and knee stabilization subtask had been implemented with dedicated controllers and tested during walking with healthy subjects. Results: Foot clearance could be increased without disrupting the subject’s own initiated walking in the remainder of the gait cycle. During weight acceptation, the knee stabilization ensured by the exoskeleton while the subject relaxed his knee. The gentle control of the robot allowed a comfortable interaction between subject and robot. Conclusion: Selective support of subtasks seems to be a viable method of interaction with the patient to train his or her gait. In the near future, we will implement a complete set of gait subtasks, which makes all kinds of training interventions possible.