Multiple Model Approach of A Soft Robotic Arm

Soft robotics is the next generation of robotics systems that complements rigid robots using concepts inspired by natural organisms. Despite possessing extraordinary flexibility, agility, and endurance, making them safe to interact with humans, modelling and controlling soft robots still prove to be a major challenge. On the one hand, data-driven model requires a huge amount of training data while accurate physics-based model suffers from computational complexity. In this paper, we present a multiple model approach to soft robotics. Here, the model is two-fold: a pressure-to-bending equation which models steady-state characteristics and several linear models to capture transient characteristics. This can be used as an alternatives to non-linear model of planar arm-shaped soft robots. We found that 2nd order exponential function achieve a good performance in modelling the steady-state bending angle while a 2nd order linear model achieve a good accuracy in predicting the transient response.