4D-Printed Soft Pneumatic Actuators Guided by Machine Learning and Finite Element Models

This paper explores the field of four-dimensional 4D-printed soft pneumatic actuators (SPAs) made completely by additive manufacturing. After manufacturing, these actuators can produce bending motions when exposed to vacuum or pressurized air stimuli. Their functionality is determined by a combination of material qualities and geometric details, which gives the pressures and motion trajectories required for fine activities like non-invasive surgery and food handling. This introduces an innovative approach to achieve four-dimensional (4D) printing of soft pneumatic actuator robots (SPAs). This method leverages nonlinear machine learning (ML) techniques combined with finite element modelling (FEM). The core of this methodology involves the development of a precise FEM that emulates experimental actuation. The primary purpose of this FEM is to generate essential training data for the subsequent ML modelling phase. The wide range of 3D printers and materials used to create pressurized air-bending-style SPAs is thoroughly examined in this paper. It also examines different approaches to modelling and regulating these actuators and provides a comparative study. This paper concludes with a summary of general considerations regarding future directions and the inherent difficulties in developing these state-of-the-art actuators