Task-based fun interaction based on rehabilitation theory promotes motor-cognitive mutual enhancement, integrating pneumatic flexible muscles and high-precision flexible sensing technology to simulate various real physical interactions. The deep combination of automatic control, Internet of Things, and artificial intelligence accurately obtains the personalized information of the trainer and matches the optimal training program, providing fun and active guided tasks and multiple training modes. It aims to help patients restore hand function, strengthen ADL, cognitive ability, and prevent Alzheimer's disease.