Design and Kinetostatic Analysis of PneuNet-Actuated Compliant Mechanisms for Soft Robots

Soft robots are challenging to design and control because they undergo large deformations and exhibit distributed compliance. This technical brief presents a simulation-based framework for soft robots constructed from PneuNet-actuated compliant mechanisms. The proposed approach integrates finite-element (FE)-generated joint data, machine-learning-based joint-response prediction, a recursive pseudo-rigid-body kinetostatic model for multijoint systems, and a forward-kinematics-based position control workflow. A revolute PneuNet-actuated cross-axis flexural (PnACF) joint is first simulated under combined pressure and external loading, and the learned joint model is then incorporated into a planar 3R robotic arm. For a representative loading case, the system-level model agrees well with full FE simulation, with errors of 7.29% in end-effector rotation, 2.40% in x-displacement, and 4.15% in y-displacement. A workspace analysis and a learned forward-kinematic model are further used for position control, allowing five selected target points to be reached with predicted error below 0.5 mm. The results demonstrate an efficient design, analysis, and control pipeline for modular soft robots.