Identification of DOE-Based Predictive Model for the Bending Behavior of a Soft Pneumatic Actuator

Collaborative robotics allows the operator to work closely with the robot without the use of barriers. In addition, completely soft end-effectors guarantee the extension of safety from the base of the robot to the tool. Soft pneumatic actuators, the most widespread soft end-effectors, present strong nonlinearities due to compressed air, large deformations, and material constitutive law. For these reasons, the identification of an analytical model is difficult, and numerical simulations are often used to evaluate the behaviour of these actuators. It is important to limit the number of simulations, which are onerous in terms of computational cost and time. In this paper, the identification of a predictive model of the bending angle of a novel soft pneumatic actuator using planned numerical simulations by the Design of Experiment method is presented. The predicted bending angles are compared with experimental results achieved on three different prototypes and with the results of a predictive formula identified in a previous work. Results show how this approach allows not only to minimize the required simulations without sacrificing accuracy but also to identify the effects of interactions between factors that would otherwise remain hidden.