Abstract: As a typical human-computer interaction system, the soft hand rehabilitation robot has the advantages of high flexibility, comfortable wearing and strong adaptability. It can help patients to carry out independent rehabilitation training, accelerate the rehabilitation process, and has been highly concerned in the field of rehabilitation therapy. However, there are still some technical challenges in software driver modeling, load analysis, intelligent perception and force control. Therefore, firstly, the key technologies of the wearable soft hand rehabilitation robot from three aspects in wearable flexible structure design, intelligent perception and intention recognition, and driving and control are elaborated. Secondly, the typical system of the latest soft hand rehabilitation robot is analyzed. Finally, the challenges of the soft hand rehabilitation robot in driver design, accuracy of intent recognition, and control algorithms are discussed. The future research trend of the soft hand rehabilitation robot is prospected.