As robots are becoming more relevant to our lives, they are still having hard time accomplishing simple tasks such as picking and lifting. Problems that include environmental constraints, pose uncertainties and hardware noises restrain robots for grasping an object successfully from a perceivable environment. Many have looked into finding best grasp candidates, reducing the degree of occlusions of current perception, and pre-manipulating object prior to grasping. However, these works focus on planning a good grasp with an assumption of precise hand placement. This thesis seeks to identify human control strategies on robot fingers by conducting human studies on grasping tasks when the object poses were constrained and noisy with respect to the robot hand. Then we elicited three finger control strategies from human data, and applied them to design simple grasping controllers to improve robot finger control at failure cases. Lastly, we framed it as a reinforcement learning problem to learn a policy that is capable at grasping a variety of objects from random graspable poses.