DEVELOPMENT OF A REHABILITATION EXERCISE FOR ASSISTED GRASPING THROUGH COLLABORATIVE ROBOT

In terms of safety, mobility, and programmability, collaborative robots constitute a progression of industrial robots. These qualities encourage the employment of these robots in the medical industry, including their use in neuromuscular rehabilitation, in addition to industrial uses. The research on robot-based rehabilitation therapy is leading to the growth of factors including the aging of the world population and technological advancements in healthcare. The rise of cobots causes increase in Human-Robot Interaction (HRI), where the functional involvement of humans has become an integral part of the design process of both robots and applications. To overcome the limitations of the traditional therapy, these years, the use of robot-assisted rehabilitation is rising. Robot-assisted therapy offers the possibility of increasing the intensity and duration of rehabilitation, enhancing the benefits of the therapy. The first cobot developed for the rehabilitation purposes as been the REHAROB. From this example, several other machines were developed, like the ARMEO robots, the MIME for upper-limb rehabilitation, the Motion Maker and the Walk Trainer. In our days industrial anthropomorphic robots such as Universal Robot (UR) robots and KUKA are employed in rehabilitation of lower and upper limbs. The aim of this study is the development of a rehabilitation exercise for assisted grasping through the UR5e robot. The exercise is based on the end-effector connection between robot and the human hand. The purpose is to reach a target with the hand, following a planned trajectory computed by the robot itself from one point to another, then grab and release the target and repeat. The project evolved following several steps. A specific handle was modeled and printed for use in the experiment. After that, starting from an exercise that was already analyzed, a control law was developed to help arm rotation during the exercise. Impedance control and a force law appended during the performance of the exercise were also developed. The exercise is divided into three modes with different difficulties, where the feature that varies is the force generated by the robot along the trajectory, which can help or resist the movement toward the target. To validate the exercise, data were acquired from ten subjects. Analyses of the data are based on the error given by the actual trajectory compared with the planned one, the characteristics expressed by the force recorded during the exercise, and the correlation coefficient between force and trajectory. From the trajectory results, it can be established that the highest mean error(±std) is produced during the easy modality (0.0428 ± 0.02 m). From the force analysis, it can be observed that the highest average human force registered by the Tool Centre Point (TCP) during the exercise is produced in medium mode with a value of 20.38± 9.71 N on the right arm. The goal of the thesis has been completed, furthermore, this work can be further pursued in the future by working on the modes of exercise.