Design and Implement of Shape Detection for the Soft Manipulator

This paper designs and implements an innovative shape detection system for the soft manipulator to attain the shape reconstruction in real time, which completes the closed-loop control system to enhance control accuracy. Cable encoders are selected as the sensor to measure the length variables for the input of the kinematics model. The kinematics model for unit section is proposed under the assumption of piecewise constant curvature theory, converting the length variables to the curve parameters. Then the unit-section model is extended to the multi-section model based on the novel D-H method, which obtains the posture of the whole manipulator. In addition, the singularity problem can be well disposed employing multivariate Taylor expansion to improve the kinematics model. Finally, we verify respectively the reasonability of the shape detection system from theory and experiment, and the experiment results agree well with the simulation results by MATLAB in real time.