Patient's Performance based Progressive Control Strategy for an Upper Limb Rehabilitation Robot

Abstract

Stroke is the leading cause of permanent disability in elders and youngsters in world. Very often, these stroke damage the motor control of an upper limb, making a victim's life very difficult in performing the daily life activities. Regaining functionality of these damaged motors is not easy. It requires many hours of laborious and resource-intensive therapeutic trainings. However, with the advancement of robotics technology, researchers have succeeded to reduce the physical and laborious work for therapists. These assisting robots were named as rehabilitation robots. Results from meta-analysis recommend that these robots might be as useful as conventional treatments. However, neither all the patients have a same level of injury nor they respond equally to a conventional “Assist-As-Needed” (AAN) algorithm, employed in existing rehabilitation robots. In this study, we have explored customized rehabilitation training strategies matching each individual's personal requirements. In particular, we have addressed challenges in rehabilitation based on patient's own medical situation and physical abilities, naming it as, “Patient's Performance based Control Strategy”. Proposed control strategy, progressively upgrades the difficulty level for the patients based on his/her participation in performing the given task. Key objective in developing the proposed strategy is to optimize the sum of kinematic error and robotic assistance in performing a given task. For this purpose, high-level control strategy is developed for planning and upgradation of a given task. Proposed control strategy carefully analyzes patient's performance for a given set of time, based on that performance, robot gently increases or decreases difficulty level for the patient. In particular, this strategy finds out the path along which patient is feeling difficulty/easiness and then decreases/increases the difficulty level specifically for that path, respectively. This is achieved by dividing the given trajectory into sub patches and analyzing the error for each sub patch. If the tracking error is more than the threshold value, difficulty level is either increased/decreased in different ways such as by increasing/decreasing the speed or range of motion or robotic assistance. This variation depends on patient's performance and past changes, the robot made. As a whole, proposed control strategy formulates a high-level control algorithm which provides not only assistance as needed but also tries to maximize patient's involvement by changing the difficulty level according to his/her abilities. To execute the planned high level strategy, low level controller is equally important. For this purpose, integral second order sliding mode control law is presented. Proposed controller is specifically designed for rehabilitation robots keeping in mind the demand of robustness against model uncertainties and external disturbances. To further enhance the robustness, for severely injured patient, disturbance observer is employed. Main reason for employing disturbance observer, is the requirement of careful tuning of controller parameters which is very tedious job for therapists. Overall system is tested with healthy subjects for simple tasks. Results show that proposed methodology finds out patient's ability and adjusts the task accordingly which concludes in maximum utilization of patient's efforts for boosting the rehabilitation recovery.