IK-MLP-based Control for a Pneumatic Soft Manipulator

Pneumatic soft manipulators exhibit a mapping relationship between the output end-effector and the input pressure, which can be obtained by inverse kinematic modelling. The traditional inverse kinematic modelling method is complex for multiple degrees of freedom (DOF) soft manipulators, especially for a pneumatic soft robot. However, in traditional inverse modelling, a Jacobian matrix needs to be solved inversely, and a singular problem will be encountered. To address this issue, this paper proposes an Inverse Kinematics-Multilayer Perceptron (IK-MLP) method for soft manipulators. In this strategy, the trained intelligent controller can be applied to control pneumatic manipulators to grasp targets without establishing a traditional inverse kinematic model. Then, based on the ground experiment system of the soft manipulator, the control algorithm is experimentally tested by using the theoretical model and the physical system. And it is proved that the given intelligent controller adopting the IK-MLP method can effectively solve the inverse singularity. The results also show that the unidirectional turning manipulator has strong controllability of 83.2% grasping ratio.