A 3D printed variable cross-section pneumatic soft manipulator with high-precision positioning capability: Design and control implementation

Soft manipulators are safer as human-robot interaction devices thanks to their inherent characteristics, such as flexibility, lightweight structure, and security. However, it is difficult to precisely control multi-segment, high-dimensional soft manipulators due to their uncertain model deformation and external disturbance. This paper develops a dual-segment variable cross-section pneumatic soft manipulator (PSM) fabricated by integrated manufacturing to solve this problem. Easier to establish inverse kinematics model (IKM), neural network IKM is trained using the PSM motion datasets. However, the neural network fitted IKM still suffers from significant position errors. In this paper, an improved control strategy combining neural network IKM and iterative feedback tuning (IFT) controller based on sensors is proposed to reduce the positioning error of the PSM tip. Two kinds of IFT laws are presented, compared, and analyzed to verify the feasibility and optimality of the control strategy through numerical simulation. By point-to-point tracking and load experiments, the method was proved to achieve 1 mm (0.27 %), which demonstrated that the improved positioning control strategy could effectively improve the accuracy of PSM under both no-load and load conditions. Finally, the dual-segment variable cross-section PSM application performance is accomplished in actual manipulation tasks.