Despite the effectiveness of soft robotic gloves in providing rehabilitation training and grip assistance for patients with hand injuries, their actuators’ lack of integrated sensors hampers precise position control, posing challenges for individuals with heightened muscular tension who are unable to achieve the desired bending angle due to this absence of feedback. In this paper, we propose a master-slave rehabilitation scheme for controlling a robotic glove using a data glove. The carbon black-based flexible strain sensor seamlessly integrates with the data glove and actuators, making it wearable and portable. The strain sensor's response was characterized in terms of strain amplitude, dynamic stability, strain rate, and fatigue characteristics. It demonstrates exceptional stretchability up to 100% with a gauge factor ranging from 1.14 to 1.34. The finger movement and grasping response were evaluated using a data glove, revealing robust capability for generating reliable control signals. The sensors were calibrated to achieve precise control of the actuator using the data glove, and a quantification was made of the relationship between finger bending angle and sensor response. A comparison was conducted between the bending angle that can be achieved by the actuator with or without feedback through establishing a pneumatic control system. The results indicate that with position feedback, maximum angle errors for thumb and index finger are 5.9° and 7.5° respectively. Furthermore, gesture recognition master-slave control experiments were carried out to provide additional evaluation on the efficacy of our proposed hand rehabilitation system.