End-effector type robots are commonly used in the rehabilitation of stroke patients because of their user-friendly and simple structure. However, as these robots are unable to measure the user's posture, a therapist is essential to supervise and guide desirable posture during exercises. Performing repetitive exercises with undesirable posture can hinder the recovery process or lead to ineffective learning. Therefore, real-time posture feedback can significantly enhance the effectiveness of training. To address this need, we have developed an algorithm for real-time posture monitoring and implemented a system capable of providing feedback to prevent undesirable postures. Our research aims to demonstrate the potential of this system in improving the outcomes for stroke patients.

  Rehabilitation robots can help stroke patients with intensive, repetitive arm movement training. However, for the training to be effective, it needs to be customized based on each patient's unique movement characteristics. This means we need a way to objectively assess how the patient's arm worked before the stroke, comparing their current performance to what is considered normal. We have developed a new method to evaluate upper limb movement after a stroke, using a model of normal arm movements. We'll use this method to create a personalized training schedule for each patient, keeping in mind the principles of motor learning. Finally, we will integrate a posture monitoring technique into this customized training plan to ensure comprehensive rehabilitation of the upper limb.