The behavior of soft robots is challenging to predict because of their many degrees of freedom and nonlinear material properties. We utilize a combination of physics-based and data-driven techniques to construct models that inform the design and control of real soft robotics systems.