My research focuses on the consequences of the embodiment of a robotic agent to generate robust behavior when interacting with the real world. Particularly dexterous soft robot hands have properties that allow for a new perspective on one of the grand challenges in robotics - autonomous in-hand manipulation. Therefore, I am investigating how to most efficiently use the morphological computation (MC) that the hand performs when physically interacting with the environment to develop new control and planning approaches. Classical analytic and learning-based approaches are not directly applicable because they were developed with a rigid hand paradigm for manipulation in mind. This motivates the investiagtion of a new set of tools to take full advantage of the body's computational power.