Data-Driven Forward and Inverse Kinematics of a Flexible Arm Based on the PCC Model

Soft robotic manipulators have great potential in multiple fields. However, the strong nonlinear motion characteristics brought about by their continuum structure make inverse kinematics a bottleneck for precise pose control. This paper proposes a hybrid modeling framework that combines a physical model with a data-driven approach. The piecewise constant curvature (PCC) model is used to analyze the forward kinematics, and deep learning is employed to solve the inverse kinematics problem. This framework takes into account both accuracy and computational efficiency and can predict the lengths of the driving cables in real time according to the target pose. The simulation results confirm the effectiveness of this framework, which can provide a new approach for the kinematic analysis of soft robots.