Shape Reconstruction of a Pneumatic Continuum Manipulator Under the Effect of Hysteresis During Trajectory Tracking

A continuum manipulator can theoretically have an infinite number of shapes for the desired end-effector pose. Taking the advantage of this behavior, this paper proposes the reconstruction of the shapes of a pneumatically actuated continuum manipulator under the effect of hysteresis to avoid collision with any obstacles present in the target trajectory. At first, a trajectory is selected from the workspace of a continuum manipulator. Subsequently, the required optimum actuation pressures for each tube of the pneumatically actuated segments are determined by using a two-way neural network model. The resulting pressure sets are then used as input to a forward kinematic model of the continuum manipulator developed based on the Cosserat-rod theory. This includes nonlinear constitutive relations based on the fractional-order Bouc–Wen model for the representation of material hysteresis behavior. The model is validated on portable bionic handling assistant, which is a pneumatically actuated continuum manipulator. The combination of the artificial intelligence model for inverse kinematics with an explicit mathematical model for forward kinematics results in greater positional accuracy in the presence of obstacles in the manipulator path.