This paper presents the design evolution of the sensing and force-feedback exoskeleton robotic (SAFER) glove with application to hand rehabilitation. The hand grasping rehabilitation system is designed to gather kinematic and force information from the human hand and then playback the motion to assist a user in common hand grasping movements, such as grasping a bottle of water. Grasping experiments were conducted where fingertip contact forces were measured by the SAFER glove. These forces were then modeled based on a machine learning approach to obtain the learned contact force distributions. Using these distributions, fingertip force trajectories were generated with a Gaussian mixture regression (GMR) method. To demonstrate the glove's effectiveness to manipulate the hand, experiments were performed using the glove to demonstrate grasping capabilities on several objects. Instead of defining a grasping force, contact force trajectories were used to control the SAFER glove in order to actuate a user's hand while carrying out a learned grasping task.