Kinematic modeling and solution of rigid-flexible and variable-diameter underwater continuous manipulator with load.

This paper presents a method to solve the kinematics of a rigid-flexible and variable-diameter continuous manipulator. The multi-segment underwater manipulator is driven by McKibben water hydraulic artificial muscle (WHAM). Considering the effect of elasticity and friction, we optimized the static mathematical model of WHAM. The kinematic model of the manipulator with load is established based on the hypothesis of piecewise constant curvature (PCC). We developed an optimization algorithm to calculate the length of the WHAMs according to the principle of minimum strain energy and obtain the configuration space parameters of the kinematic model. Based on the infinitesimal method, the homogeneous transformation matrices of the variable-diameter bending sections are computed, and the terminal position and attitude are obtained. In this paper, we studied the working space of the manipulator by quantitative analysis of the impact factors including pressure and load. A deep neural network (DNN) with six hidden layers is designed to solve inverse kinematics. The forward kinematic results are used to train and test the DNN, and the correlation coefficient between the output and target samples reaches 0.945. We carried out an underwater experiment and verified the effectiveness of the kinematic modeling and solution method. [ABSTRACT FROM AUTHOR]