| Summary: | Robots are increasingly utilized in upper limb physical rehabilitation to intensify practice and reduce therapists' burden by minimizing the need for manual patient assistance. While robotic platforms have gained prominence in clinical physical therapy, their effectiveness in translating into daily activities remains limited. Research indicates that most stroke survivors continue to require assistance with upper limb tasks after intensive therapy. This gap underscores the need for a versatile upper limb assistive robot that is compact, easy to use, adaptable to various activities of daily living (ADL), and cost-effective.
This thesis introduces the Assistive Robotic Arm Extender (ARAE), a compact 3D end-effector-type robot designed for the clinical settings. The ARAE aims to assist patients in performing ADLs and engaging with real objects, enhancing their independence. We hypothesize that the ARAE will enable patients to initiate and perform ADL tasks independently, improving muscular effects and abnormal postures, without altering the smoothness of motion.
To achieve the stated objectives and hypotheses, this study first developed a robotic system, including both the mechanical and software components, to assist human upper limb movement in 3D space. The ARAE utilizes a kinematic model that enables point-to-point control within the designated task space. Furthermore, the system is designed to compensate for the robot’s self-weight. The interactive workspace has been carefully designed to accommodate the range of motion required for Activities of Daily Living (ADLs) tasks by 95\% of the population, thereby ensuring broad usability.
Then we introduced an adaptive arm support control scheme designed to compensate for the gravitational force in three-dimensional space on the upper limb, based on the posture of the human arm. This innovative control scheme comprises two primary components: 1) two models for estimating human joint angles that do not require additional wearable sensors, and 2) the calculation of support force based on these estimated joint angles. Our experiments demonstrated that the sagittal plane model substantially improves angle estimation accuracy within the sagittal plane, especially when adapting to significant torso movements, highlighting the model's adaptability and precision. Subsequently, we evaluated the effects of the assistive force on healthy subjects, confirming that our framework significantly reduces muscular effort. Specifically, the muscular activity of the Biceps Brachii decreased by an average of 66.97% during a hand-to-mouth task, thereby validating the effectiveness of our proposed control scheme.
Finally, we carried out a feasibility clinical trial with three chronic stroke patients performing ADLs. Patients wore robots and performed ADL tasks in two conditions: with assistance from the robot (With Robot) and without assistance (No Robot). The performance of the ARAE system was assessed using physiological and kinematic metrics, including muscular efforts, motion smoothness, and the kinematics of human movements. The device's high transparency ensures it integrates seamlessly with natural movement patterns, thus enhancing user comfort and task performance. Notably, the ARAE system significantly reduces muscle activity, offering particular benefits to patients with low muscle tone. However, challenges persist for those with high spasticity due to complex interactions and increased resistance from muscle tone. Furthermore, the ARAE effectively improves kinematic outcomes, notably increasing shoulder flexion and external rotation, thus compensating for abnormal movements and broadening the range of achievable motions during ADL tasks. Feedback from questionnaires completed by participants yielded both positive and constructive responses. The findings of this thesis demonstrate the substantial potential of the ARAE as an assistive device for chronic stroke patients with upper limb disabilities, supporting them in performing Activities of Daily Living (ADLs) within clinical settings.
|
|---|